I assume that we are talking about some sort of statistical analysis in which different factors are assigned different weights. What result you get then depends on how you weight the individual factors, and downweighting some of them will produce a different outcome.

There are plenty of uses of the term in statistical texts, e.g.
“If a generalized least-squares regression method (such as the IGRS (WRS) method, as in (6.18) and (6.19) above is used to correct for these unequal error variances, in effect downweighting the longer-time-interval repeat-sales pairs in the regression, then there will a tendency for standard errors to be larger on longer-interval index differences. Even so, the effect of such downweighting in causing unequal standard errors may be much less than one might think.”
https://books.google.co.uk/books?isbn=0191521655