RESUMEN
Previous studies have indicated that the presence of atypical end points (e.g., 1,639 and 2,897) on a number line has a negative effect on number line estimation (NLE) performance (Booth & Newton, 2012; Hurst, Leigh Monahan, Heller, & Cordes, 2014). In the present study, we investigated whether this effect could be attributed to a disruption in the ease with which benchmarks on the number line can be determined and whether this possible disruption changes with age. Hence, we asked 5th graders and adults to perform a NLE task in a typical and an atypical condition with a number line ranging from 0 to 1,000 and from 367 to 1,367, respectively. Results showed that participants' overall estimates were less accurate in the atypical condition compared to the typical condition. A similar pattern of results was observed for the estimates around the self-created benchmarks at 50% in children and at 25%, 50%, and 75% in adults. Moreover, in both age groups, we found a shift from a larger proportion of estimation patterns' being best fit by more complex power models in the typical condition toward a larger proportion of estimation patterns best described by simpler power models in the atypical condition. Of importance, we also found evidence that adults' estimation patterns in the typical condition are best described by a 4-cycle power model, hereby extending previous modeling results. Overall, our findings indicate that hindering the application of benchmark-based strategies negatively affects children's and adults' NLE performance. (PsycINFO Database Record (c) 2018 APA, all rights reserved).