Every form of testing and measurement relies on the ability to trust the mechanism or process that assesses the outcome of a test. For example, when you take a package to a shipping company you rely on its scale for an accurate weight that results in an appropriate shipping fee. If the scale registers the wrong weight -- the same or a different incorrect measurement each time -- you're dealing with the forms of error referred to as "unsystematic" or random and "systematic" or predictable.
Systematic errors produce predictably incorrect results from a measuring process. Suppose your package weighs 15 pounds on your personal scale, but the shipping clerk shows you that the package actually weighs 10 pounds. If you later discover that your scale read 15 pounds every time you weighed the package because you calibrated your scale improperly, your scale introduced a systematic error because the equipment operated predictably but incorrectly.
Unsystematic errors constitute random variations. If you weighed your package with your hand resting on it, the pressure of your hand incorrectly inflated the weight of your package. Each time you weigh the package, the outcome varies depending on how hard you press on the scale. Your unsystematic error alters the outcome of the measurement, producing a random outcome each time because of variation in pressure.
Sources of Error
The sources of unsystematic error vary depending on the types of measurements they affect. Their common ground stems from their random impact on results. If four teachers administer the same test and score the answers differently, they introduce unsystematic error into the results. If a piece of equipment operates at the wrong speed because of fluctuations in electrical power, its output varies based on unsystematic error. Identifying and eliminating unsystematic errors requires eliminating the prospect that a systematic error influences results. You can identify a systematic error fairly easily by looking for predictably incorrect results.
Assumptions allow error to creep into testing results. For example, suppose you state a simple mathematical problem using highly complex sentences that require advanced language ability. If you use this math problem to test people's understanding of its underlying simple arithmetic, you actually may be testing language skills and not math competence. Your assumption introduces unsystematic error into your results. To screen out random variation in results, always examine your assumptions first.
- Journal of Economic Psychology: Random Error in Willingness to Pay Measurement: A Multiple Indicators, Latent Variable Approach to the Reliability of Contingent Values
- Inferential Statistics: Basic Concepts; W. K. Gabrenya Jr.
- U.S. Office of Special Education Programs: Reliability Issues and Evidence
- Reliability and Validity of Measurement; Garry Chick
- Appendix E: Introduction to the Scientific Method; Frank L. H. Wolfs
- Experiment-Resources.com: Systematic Error
- Purdue University Division of Chemistry Education: Bodner Research Web: Errors
- Southeastern Louisiana University: Random Error and Systematic Error
- Photo Credit Thomas Northcut/Photodisc/Getty Images