Measuring and Correcting Response Heaping Arising From the Use of Prototypes

Abstract

Imprecision in respondent recall can cause response heaping in frequency data for particular values (e.g., 5, 10, 15). In human dimensions research, heaping can occur for variables such as days of participation (e.g., hunting, fishing), animals/fish harvested, or money spent on licenses. Distributions with heaps can bias population estimates because the means and totals can be inflated or deflated. Because bias can result in poor management decisions, determining if the bias is large enough to matter is important. This note introduces the logic and flow of a deheaping program that estimates bias in means and totals when people use approximate responses (i.e., prototypes). The program can make estimates even when spikes occur due to bag limits. The program is available online, and smooths heaps at multiples of 5 (numbers ending in 5 and 0) and 7 (e.g., 7, 14, 21), and produces standard deviations in estimates.