generalizability
jen-er-al-ize-uh-BIL-uh-tee
Definition
The extent to which the results of a study or statistical model can be applied or remain valid when tested on different populations, settings, or new sets of data.
Etymology
Derived from the Latin 'generalis,' meaning 'relating to a whole class,' combined with the suffix '-ize' and '-ability.' It entered English to describe the quality of being able to be extended or applied more broadly.
In the news
The article uses this term to describe a study that tested whether existing psychological models actually work when applied to a new, independent sample of couples. It highlights that a model's effectiveness in its original study does not guarantee generalizability to the wider population.
A divorce-prediction equation was 90 percent accurate on the data it was built from, but when it flagged a couple as heading for divorce in a fresh sample it was right only 29 percent of the time, which is the trouble with rules like these
Read the full article ↗ScienceBlog.com