Search

Convergence rates of cluster analysis results are studied for six sets of partition sorting data. Hierarchical cluster analysis, fuzzy cluster analysis, k-medoids clustering and four variants of the consensus clustering method are covered. The six data sets are kinship terms, colour plates, drug data, consultancy Web site items, social groups and prenatal tests. Convergence rates depend on the number of clusters extracted, on the clustering method and on the domain. For domains of 25 items or more, cluster analysis results identical to the final result were most consistently obtained using the third method of Gordon and Vichi (2001). Example evaluations are given for the most stable numbers of clusters for five sets of partition sorting using the method of Gordon and Vichi. The interpretations resemble previously published interpretations of results obtained from non-partitioning statistical methods.

Application supplémentaire de l'analyse de fréquence configurale pour le tri libre des données :L'intention de cet article est de démarrer une discussion à propos de cet outil simple et supplémentaire pour analyser des tris libres de données. Des caractéristiques communes des tris de données et de leur pertinence pour les analyses de fréquence configurale sont discutées, en relations avec l'analyse des données descriptive et confirmatoire. La taille d'échantillon pour le tri de cartes est re-examinée pour l'analyse descriptive. L'échantillon nécessaire dépend d'une certaine mesure du domaine spécifié. Le nombre moyen de différentes configurations rencontrées dépend du nombre de tris recueillies, le nombre d'éléments inclus dans le domaine et les caractéristiques du domaine. Un test simple et exact de signification est proposé pour tester les hypothèses de fréquences configurales de tri libre et flou de données.This paper analyses the application of Configural Frequency Analysis (CFA) to free sorting data. It is intended to start a discussion about this simple supplementary analysis tool for free sorting data. Common features of sorting data and their relevance for CFA are discussed, in its approach to using both descriptive and confirmatory data analysis. Sample size requirements of card sorting are also revisited for descriptive analysis. The sample required to some extent depends on the specified domain. The average number of different configurations encountered depends on the number of sortings collected, the number of items included in the domain and the qualities of the domain. A simple exact significance test is proposed for hypothesis testing of configural frequencies in free fuzzy sorting data.

Abstract.Mental models of domains may be hierarchical in nature and may include several levels of abstraction. In this article, initial emphasis is placed on reviewing past and current methods and practices in card sorting with respect to their suitability to capture such mental models. Then a new variant of card sorting is described that improves the collection and reconstruction of an individual’s semantic tree model of a domain. Thus, this variant allows sortings to be performed that use a truly free ordinal scale as well as an interval scale. The mathematical processes underlying the method are briefly described. Finally, practical examples of the new variant in card sorting are presented and the findings are discussed.

Zur stratigraphischen Einstufung des besonders gut aufgeschlossenen Unterkreide-Profils südlich der Erzherzog-Johann-Klause (Thierseemulde, Nördliche Kalkalpen) konnten mit Hilfe von Foraminiferen und Ammoniten verbesserte Daten gewonnen werden. Auch hier tritt wie im östlich gelegenen Glemm-Bach (Weidich 1990b) eine für kalkalpine Verhältnisse artenreiche, Nodosariiden- und Epistomina-dominierte Residualfauna auf. Bemerkenswerterweise kommt die Gattung Leupoldina auch an dieser Lokalität innerhalb der Nördlichen Kalkalpen vor.