Number & MeasureGreek → English
Homoscedasticity
/ˌhoʊmoʊskɪdæsˈtɪsɪti/noun
The condition in which a variable's errors have constant spread across all values of the predictors.
The Greek roots
ὁμός
homós
same, shared
σκεδαστικός
skedastikós
able to scatter
Literally: “same scattering”
The story of the word
Pearson made the word in 1905 as the well-behaved twin of heteroscedasticity, from σκεδάννυμι, to scatter. It is one of the few statistical terms that names an assumption rather than a finding: ordinary least squares takes it for granted, and most of the diagnostics in a regression textbook exist to check whether the assumption held. The first root is the ὁμός of homogeneous, not the Latin homo of human.
- How it travelled
- Ancient Greek → scientific coinage → English
- First recorded
- 1905
Same family
heteroscedasticityhomogeneoushomonymanomaly
In a sentence
The test came back clean, so homoscedasticity was one thing they did not have to argue about.