Digit Preference in African Survey Data and Their Impact on Parametric Estimates
No Thumbnail Available
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Description
Most microeconomic and demographic variables in African countries are
collected from sample surveys. Some are comprehensive and cover the whole country
while others are region or area specific. In many large scale surveys an appropriate
scientific method of sampling (usually multistage stratified cluster sampling) is adopted.
However the responses may not be accurate. There are many reasons for wrong reporting
such as memory lapse or deliberate attempt to underestimate (such as income) or
overestimate (such as expenditure). The most common source of error is the tendency to
provide numerical responses that end with certain digits, especially those that end with
integer “zero”, followed by the integer “five”. A typical case is “age heaping” where
individuals give their ages with umbers ending with these digits. The types of digit
preference are not only limited to age reporting. When farmers are asked about acreage
planted, amount harvested, number of cattle owned, output consumed and sold , distance
from home to the nearest market etc., they appear to give numerical values that end with
“zero” or “five”