NotesFAQContact Us
Search Tips
ERIC Number: ED064402
Record Type: RIE
Publication Date: N/A
Pages: 17
Abstractor: N/A
Reference Count: 0
On the Analysis of Partitioned Data.
Pruzek, Robert M.; And Others
A description is given of a general method for studying partitions. The main focus is with the analysis of relationships among several different partitions of the same items for the explorations as well as confirmation of structural relationships. A partition is defined as a set of mutually exclusive clusters of items; however, this paper deals with overlapping or conjoint clusters also. The terms "clustering" and "partitioning" are found in the context of both data collection and analysis. The major impetus for the present method was the problem of studying how an individual (manifest) partition might be examined in relation to a single derived (latent) partition in the context of latent partition analysis (LPA). In the process of studying characteristics of the distribution of the principal statistic, there was also developed an approach to the organization of a set of partitioned data with respect to any specified target. Data were analyzed first with respect to an a priori target which had been generated by the investigators on the basis of characteristics of the Morse Code symbols, with no regard for the letters themselves. It was found that: (1) The organization and display features of the proposed strategy may be of greatest significance, especially when stimuli or items are complex and when sorters are heterogeneous in terms of how they formed their categories; and (2) There are several other possible indices of association for comparing matrices. (CK)
Publication Type: N/A
Education Level: N/A
Audience: N/A
Language: N/A
Sponsor: N/A
Authoring Institution: N/A