March 2006:  adjustments to the MDO, and support for "did-not-see" cases

 

Pairwise exclusions now apply to the calculation of cognitive item correlations when MDO is used on the *sub card: all item-criterion correlations are now corrected for missing data.

 

A did-not-see option has been added to the System worksheet. This option works in a manner analogous to MDO: if a person has not been presented with the chance to answer an item (for whatever reason), the calculation of item statistics is adjusted accordingly. A gentle click here will let you see more.

 

Both of these adjustments will be of use when students see different test items. In some current online testing systems (for example), items are sampled from an item bank, with each student getting a subset of items. Not only do students see a sample of test items, but the items included in the sample will vary from student to student, effectively presenting each student with a different test.