In scientific research, measurement error or observation error is the difference between an observed value, that is the result of measurement of something and the true value of what we are measuring. It is also known as experimental error.
There are two common types of measurement error:
It is the chance difference between the observed and true values of something. For example, when a researcher misreads a weighing scale and records an incorrect measurement, it will be a random error.
It is the consistent or proportional difference between the observed and true values of something. A miscalibrated scale consistently registering weights as higher than they actually are, is an example of systematic error.
Different measures of error include absolute error, instrument error, greatest possible error, margin of error, percent error, relative error among others.
By recognizing the error sources, researchers can reduce their impacts and record accurate and precise measurements. When gone unnoticed, these errors can lead to research biases like information bias or omitted variable bias. Statistical procedures like standard error of measurement, coefficient of measurement etc can be used to assess absolute reliability.
List of recommended resources #
For a broad overview #
The article by Matthew Blackwell, James Honaker and Gary King presents an easy-to-use framework for treating missing data problems in cases of measurement error.
This broad overview studies survey responses and uses data collected from them to develop measurement error bounds on treatment effects estimated from surveys.
For in depth understanding #
Edited by Paul J. Lavrakas, In conjunction with top survey researchers around the world and with Nielsen Media Research serving as the corporate sponsor, the Encyclopedia of Survey Research Methods presents the latest information and methodological examples from the field of survey research.
This book by Madhu Viswanathan is motivated by the lack of literature that enhances our understanding of measurement error, its sources, and its effects on responses. The purpose of the book is to enhance the design of research, both of measures and of methods.
Case study #
This research paper by Paul Glewwe examines ways of reducing or removing measurement error in income and expenditure data collected from household surveys using instrumental variable methods.
This paper co-written by John Gibson, Kathleen Beegle, Joachim De Weerdt and Jed Friedman compiles data from eight different consumption questionnaires randomly assigned to 4,000 households in Tanzania to obtain evidence on the nature of measurement errors in estimates of household consumption.
The blog gives an overview of measurement error in statistics along with several ways to reduce measurement error.
This blog post provides a brief overview of the different types of measurement errors in statistical analysis.