Hierarchical Cluster Analysis Options

This dialog selects output to be produced from a hierarchical cluster analysis.

Display

Dendrogram	The dendrogram is a hierarchical tree showing how the clusters are related.
Amalgamations	Lists the order in which clusters are merged, with the corresponding similarity (not available for single link/nearest neighbour method – this information is contained in the minimum spanning tree for that method)
Minimum spanning tree	This is a tree connecting all n data points, such that the links have minimum total length
Nearest neighbours	Lists the nearest neighbours of each data point
Typical elements	List the average similarity each point has with other points in the same group. Within each group, the point with the highest value can be considered most typical
Mean similarities	Displays mean similarities between and within groups.

Plot ordination and minimum spanning tree

An ordination is obtained from the similarity matrix using principal coordinates analysis and the first two dimensions are plotted, with the minimum spanning tree superimposed.

Plot dendrogram

A dendrogram is plotted to show the hierarchical relationships between the units, which is ordered according to the results of the cluster analysis.

Display unit labels

If selected, this enables a dropdown list in which you can specify unit labels to be used on the dendrogram. If you do not make a selection then the row labels from the similarity matrix will be used for the labels on the dendrogram.

Display similarity axis

If selected, an axis for the similarities is displayed on the graph.

Orientation

This controls the orientation of the dendrogram:

Horizontal	The units are displayed on the left of the graph with the links running horizontally across the page
Vertical	The units are displayed on the bottom of the graph with the links running vertically down the page

Printing threshold

Specifies, as a percentage, how often clustering information will be output. For example, setting this to 10, will print out clusters formed at 90%, 80% and so on.

Forming groups

Specifies, as a percentage, at what level of similarity groups are to be defined from the current set of clusters, and saved for displaying Typical elements or Mean similarities.