The data are entirely fictitious as are its labels. The original labels were even different from the present ones. The data were used to demonstrate the principles behind the Hiclas model (hierarchical classes models). It is assumed that data have been collected data from four subjects (Eva, Katrijn, Iven, Paul) whose present suitability for four jobs (university administrator, full professor, assistant professor,post-doc) has been assessed by seven raters (A--G)
A three-way data array X = (x(i,j,k)) has the following form
|-----|i=1 |-----| |i=2 |-----| | |.. | | | |.. | | |____|i=I=7 k=K=4 | |____| k=2 |_____| k=1 j=1,.,J=4
The actual data file has the following form:
j=1,.,J=4 |-----|i=1 | |i=2 | |.. k= 1 | |.. |_____|i=I=7 |-----|i=1 | |i=2 | |.. k= 2 | |.. |_____|i=I=7 |-----|i=1 | |i=2 | |.. k=4 | |.. |_____|i=I=7
Thus the first mode (i) is nested in the third mode (k) and there are 7 (Raters) times 4 (University Positions) rows and 4 (Persons) columns.
Hiclas models for binary data do not require preprocessing. For other three-mode models these data should probably be treated as three-way rating data. For details see Kroonenberg (2008). Applied multiway data analysis. Hoboken NJ: Wiley (Chapters 6 and 18).
[Download the zipped University Positions Data]