Nonparametric Partial Correlation

Abstract

INTRODUCTIONIt is often desired to measure the correlation between two variables, say X and Y, controlled for a third variable, say Z. Any such measure may be called a pallt,i,a.lc.oMei.a.tion, written C(X,YIZ).Here .Q. indicates c.oMe-~on and l indicates c.ontJLolied 60~; the varying interpretations of these two concepts form a basis for distinguishing among the indices proposed so far in the literature.In the next three sections I shall briefly review •these concepts and the indices derived from them; after.wardsI shall present a new index with examples and discussion.2. CONCEPTS• OF CORRELATION Consider first correlation between X and Y, with Z being ignored at present.In general terms one may say that X and Y are positively correlated if there is a tendency for high values of X to occur together with high values of Y, and low values of X with low values of Y; they are negatively correlated if high values of X tend to occur with low values of Y, and low X with high Y.And correlation p~ ~e means either posi~ive or negative correlation.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 38 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.