semi-partial
UK[ˌsɛm.iˈpɑː.ʃəl]US[ˌsɛm.iˈpɑːr.ʃəl]
adj
(statistics) describing a correlation between one variable and another, from which the effects of one or more other variables have been removed from the second variable but not the first.
Morpheme Breakdown
semi
partial
semi
half
partial
incomplete
Etymology
The term 'semi-partial' is a modern English compound, primarily used in the specialized field of statistics. Its first element, 'semi-', is a direct borrowing from Latin, where 'semi' meant 'half', and it entered English as a productive prefix to indicate something is partly, but not fully, the case. The second element, 'partial', has a longer journey, originating from the Latin 'pars' (a part, piece, share), which derives from the Proto-Indo-European root pere- (to grant, allot). The adjective 'partial' thus literally means 'of or pertaining to a part'. In statistics, the compound 'semi-partial' logically describes a specific type of correlation that is 'partly partial', where the adjustment for other variables is applied to only one of the two variables in the correlation pair, hence a 'half-adjusted' partial correlation.
Analysis
Structure: semi (half) + partial (incomplete)
semi: Latin origin, meaning 'half'. In English, it functions as a prefix denoting incompleteness or partiality.
partial: From Latin 'partialis', from 'pars' (part). In English, it functions as an adjective meaning relating to or being a part, not total, or biased.
Examples
The researcher reported a semi-partial correlation to isolate the unique contribution of the predictor.
Understanding the difference between partial and semi-partial correlation is crucial for advanced regression analysis.
The semi-partial correlation coefficient was small but statistically significant.