Ceiling effect (statistics)

Ceiling effect (english ceiling effect) refers to a measurement error that is due to the exceeding of the measuring range. Its counterpart, a measurement error due to underflow of the measuring range, ie ground effect (english floor effect).

Example

If you. A letter scale indicating a maximum of 100 grams, hangs a letter that weighs about 100 grams, or a package containing several kilograms, the scales always shows 100 grams, regardless of how severe the issued subject is actually

Test theory

A good test to determine the differences between the test objects, even on the " edges ", ie at the very strong and very weak test takers. If a test so simple that many subjects optimum performance (eg, the maximum score ), although their capabilities are very different, one speaks of the ceiling effect. For example, if an intelligence test is designed so that it can not reliably measure the intelligence quotient ( IQ ) of a person from a certain limit, one speaks of the ceiling effect. In general, this limit is about 140: That is, the correct completion of each question in the intelligence test leads to an IQ of 140 An even better result, the test can not deliver, no matter how intelligent the subject really is.

• Example:

Consider a fictitious intelligence test, maintain its creator, so to be able to reliably cover the IQ range of 50 to 150 IQ points. The test consists only of several mathematical subtests in which the subject is asked to continue according to logical criteria given series of numbers. The most demanding series of numbers may have the following form:

A) 1 2 3 4 5 (solution: 6) b ) 2 4 6 8 10 (solution: 12) c) 3 6 9 12 15 ( Solution: 18), etc.

Since these tasks can be solved by many, even intelligence weaker test takers correctly probably, but intelligence strong subjects should certainly have no problems, probably a high score by working through all the tasks is achieved by almost all subjects. The test results of most thus reach the "ceiling " ( engl. ceiling ), and the desired significance takes damage.

In the opposite case, ie if the test is so difficult that many subjects zero or a few score points, although their performance is different, this is called ground effect and floor effect.