Signifikanz, Effektstärke und Konfidenzintervall

Autor:	Hermann Faller
Rok vydání:	2004
Předmět:	education.field_of_study Sample size determination Sample (material) Statistical significance Rehabilitation Statistics Population Probability distribution education Null hypothesis Confidence interval Mathematics Term (time)
Zdroj:	Die Rehabilitation. 43:174-178
ISSN:	1439-1309 0034-3536
DOI:	10.1055/s-2003-814934
Popis:	The term "statistical significance" is often misunderstood. The result of a study may be labelled to be "highly significant" as if implying "highly important". Statistically significant, however, does only mean that a study result might have been found with a predefined probability (conventionally set at 5 %) even when the null hypothesis is true in the population, i. e. the effect found in the study sample does not exist in reality. Whether a result proves to be significant or not largely depends on sample size. Thus, in a large sample minimal effects of no practical relevance may turn out significant whereas in a small sample even large, important effects may fail to reach the significance level. As a consequence, when presenting the results of a study the effect size should be reported together with a confidence interval indicating the probable range that contains the population effect.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::125e704efde152a1f1dc089cf970468d https://doi.org/10.1055/s-2003-814934 Zobrazit plný text záznamu