right now, the default behavior for all functions is that sorts are, in fact interval scaled.
this should be explicated in the docs.
This makes sense, because interval scaled data can actually be used in an ipsative fashion, and, anyway, most analysis relies on interval scaled data (PCA etal).
If ever there were a truly ordinal treatment of sorts, this should also be reflected in the data and methods.
Example method for this is the psClosedSorts -> psSort coercion: with interval data, any missing integers (say, no item at -3) must be replaced. With ordinal interpretation, this would be unecessary, actually, empty columns would be altogether meaningless and could be omitted.
Will need to think quite carefully about where this plays out.
right now, the default behavior for all functions is that sorts are, in fact interval scaled.
this should be explicated in the docs.
This makes sense, because interval scaled data can actually be used in an ipsative fashion, and, anyway, most analysis relies on interval scaled data (PCA etal).
If ever there were a truly ordinal treatment of sorts, this should also be reflected in the data and methods.
Example method for this is the
psClosedSorts
->psSort
coercion: with interval data, any missing integers (say, no item at-3
) must be replaced. With ordinal interpretation, this would be unecessary, actually, empty columns would be altogether meaningless and could be omitted.Will need to think quite carefully about where this plays out.