G * Power
G * Power
|
|
---|---|
G * Power version 3.1.9.2, 2014 |
|
Basic data
|
|
developer | Axel Buchner, Edgar Erdfelder, Franz Faul, Albert-Georg Lang |
Current version | 3.1 (2014) |
operating system | Windows, Macintosh |
programming language | C ++ |
category | Statistics software |
License | Proprietary, free of charge |
German speaking | No |
G * Power (HHU Düsseldorf) |
G * Power is a computer program for carrying out statistical tests , especially power tests. G * Power was developed by four German psychologists ( Axel Buchner , Edgar Erdfelder , Franz Faul, Albert-Georg Lang) since 1992. The program is protected by copyright, but free for everyone.
Range of functions
Unlike most programs for statistical analysis of empirical data ( statistical software ), G * Power does not work with the raw data, but provides routines for various tests on an aggregated level (characteristic values and distributions). Therefore, there is no data matrix in G * Power that records the data for each case and each variable. G * Power offers approximately 50 different tests, which are divided into five distribution-based groups (referred to in the program as "Test family"):
- "Exact": tests on any probability distribution based (eg. Binomial test , McNemar test , Fisher's exact test )
- "F tests": tests on the F distribution are based,
- "t tests": tests based on the t-distribution based,
- "χ² tests": tests based on the chi-square distribution ,
- "z tests": tests based on the normal distribution .
There are five analysis types (referred to in the program as "Type of power analysis") available for each test:
- "A priori": Calculates the required sample size if the error level for the error type 1 (α), the power (1-β) and the effect size are specified,
- "Compromise": Calculates the implicit α and the power, if the ratio of α to β, the sample size and the effect size are given,
- "Criterion": Calculates α if power, effect size and sample size are given,
- "Post hoc": Calculates the power achieved if α, sample size and effect size are specified,
- "Sensitivity": Calculates the effect size if α, power and sample size are specified.
All tests are available for one-sided and two-sided viewing. For individual parameters required as input, it is possible to determine these from other parameters.
Versions and operation
The first published version was called G * Power 2. In 2007, G * Power 3 was published with an extended range of functions (including the addition of the test types "Compromise" and "Criterion"). The operation is completely menu-driven. The user interface is only available in English.
License
G * Power is protected by copyright by the authors (so it is not “ free software ”, although “free for everyone”). The program is available free of charge on a website of Heinrich Heine University Düsseldorf , commercial distribution is expressly prohibited, screenshots may be used without permission.
literature
- E. Erdfelder, F. Faul, A. Buchner: GPOWER: A general power analysis program. Behavior Research Methods. In: Instruments, & Computers. Vol. 28, 1996, pp. 1-11.
- A. Buchner, E. Erdfelder, F. Faul: Test strength analyzes. In: Manual of Quantitative Methods. 1996, pp. 123-136. doi: 10.25521 / hqm10
- F. Faul, E. Erdfelder, A. Lang, A. Buchner: G * Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. In: Behavioral Research Methods. Vol. 39, no. 2, 2007, pp. 175-191.
Individual evidence
- ^ E. Erdfelder, F. Faul, A. Buchner: GPOWER: A general power analysis program. Behavior Research Methods. In: Instruments, & Computers. Vol. 28, 1996, pp. 1-11.
- ^ F. Faul, E. Erdfelder, A. Lang, A. Buchner: G * Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. In: Behavioral Research Methods. Vol. 39, no. 2, 2007, pp. 175-191.