We investigated whether three interventions -- studying incorrect worked examples, studying correct worked examples, or receiving feedback -- improved children's 0-1,000 (Experiment 1) and adults' 1 thousand--1 billion (Experiment 2) number-line estimation precision relative to a no intervention control group. At pretest, participants estimated numbers on number lines (without feedback or examples) and judged their confidence in each estimate. They were then assigned to an intervention group, completed intervention trials, and then estimated numbers and judged their confidence at posttest. Children's and adults' confidence judgments did not increase from pretest to posttest. However, consistent with prior research, feedback improved children's and adults' estimation precision from pretest-to-posttest. Importantly, studying correct worked examples also improved precision, but studying incorrect worked examples did not. Although not all children benefited from training, those who did were able to transfer their knowledge to a number-categorization task. Adults in the feedback group were also able to transfer their learning. These outcomes highlight benefits of studying worked examples for improving children's and adults' estimation precision, but to obtain benefits, corrective feedback must be included.