Download full text
(7.553Mb)
Citation Suggestion
Please use the following Persistent Identifier (PID) to cite this document:
https://nbn-resolving.org/urn:nbn:de:0168-ssoar-34309
Exports for your reference manager
Einführung in die Clusteranalyse mit SPSS-X für Historiker und Sozialwissenschaftler
Introduction to cluster analysis using SPSS-X for historians and social scientists
[journal article]
Abstract In dem vorliegenden Skript werden unter den Verfahren der Clusteranalyse das 'hierarchisch-agglomerative Verfahren' und ein 'allokatives Verfahren' behandelt. Das Schwergewicht liegt auf der konkreten Umsetzung und technischen Realisierung methodologischer Regeln anhand von Übungen im Statistikprogr... view more
In dem vorliegenden Skript werden unter den Verfahren der Clusteranalyse das 'hierarchisch-agglomerative Verfahren' und ein 'allokatives Verfahren' behandelt. Das Schwergewicht liegt auf der konkreten Umsetzung und technischen Realisierung methodologischer Regeln anhand von Übungen im Statistikprogrammpaket SPSS-X. In zwei weiteren Kapiteln werden Verfahren der Stabilitätsprüfung und Verfahren der Behandlung fehlender Werte diskutiert. Die theoretisch vorgestellten Klassifikationsprobleme werden anhand eines Beispiels durchgerechnet. Das Ziel der vorgestellten Klassifikation besteht darin, familiale Haushaltstrukturen in einer osttiroler Gemeinde am Ende des 18. Jahrhunderts zu bestimmen. (pmb)... view less
'This article is addressed to users of classification procedures in the social historical sciences. According to this aim an example from historical family research is used to describe the steps necessary to solve a classification task. These steps are: (1) Selection of classification attributes and... view more
'This article is addressed to users of classification procedures in the social historical sciences. According to this aim an example from historical family research is used to describe the steps necessary to solve a classification task. These steps are: (1) Selection of classification attributes and units. (2) Treatment of missing data. (3) Transformation of classification attributes to comparable scales. (4) Standardization of classification units. (5) Selection of dissimilarity and similarity measures. (6) Selection of classification procedures. (7) Calculation of cluster solutions. (8) Validition of cluster solutions by stability and sensitivity analysis. As can be seen from the previous list some steps - especially step (2), (3) and (8) - are neglected or underestimated in most books on cluster analysis, although they are of practical importance: How can missing data be treated? What are the effects of different treatments of missing data on classification results? Is it better to transform classification attributes to comparable scales by empirical or theoretical procedures? How do these different methods of data transformation influence the results of cluster analysis? Finally, how can the validity of a cluster analysis be tested? The article tries to answer these questions. Furthermore standard text books on cluster analysis pay little attention, how a user of statistical program packages can realize methodological rules within the program used: How can certain types of dissimilarity measures be calculated without specific option in the program used? How can data transformation be realized? Or, how can a sensitivity analysis be performed, when there is no specific program to do this? In the article the statistical program package SPSS-X is used to demonstrate the realization of methodological rules. This investigation shows, that a wide variety of methodological rules can be realized within SPSS-X, if the user writes small programs. However there are certain limitations, expecially to the treatment of missing data. Exercises complete the represantation of the single steps. They can be solved without any computer.' (author's abstract)... view less
Keywords
electronic data processing; statistical analysis; housekeeping; classification; methodology; Austria; cluster analysis; eighteenth century; empirical social research; historical social research; family
Classification
Methods and Techniques of Data Collection and Data Analysis, Statistical Methods, Computer Methods
Document language
German
Publication Year
1989
Page/Pages
p. 6-167
Journal
Historical Social Research, 14 (1989) 2
DOI
https://doi.org/10.12759/hsr.14.1989.2.6-167
ISSN
0172-6404
Status
Published Version; peer reviewed