[Rasch] Repeated observations
s-luppescu at uchicago.edu
Fri Oct 12 02:37:38 EST 2007
Hello, I have a data set in which individuals were observed on multiple
occasions, and rated on a number of criteria. In some cases, a
particular practice was not observed (or was ignored) and was not rated.
These people would have fewer data. In addition, some individuals were
just observed more frequently than others, and so have more data.
The question is, how to deal with these multiple observations. They can
not be treated as different items, because the order is not meaningful
or consistent across individuals. I am tempted to use the data to get
multiple measures for people and then average them (weighting by the
precision) to get a single person measure.
Somehow it seems that I could use facets for this while ignoring the
rater facet (there is not overlap in the ratings -- only one rater ever
rated one other person) but I'm not sure how I would do it.
Anyone have any creative solutions?
Stuart Luppescu -=- slu .at. ccsr.uchicago.edu
University of Chicago -=- CCSR
才文と智奈美の父 -=- Kernel 2.6.20-gentoo-r7
Acceptance testing: An unsuccessful attempt to
More information about the Rasch