Online data analysis

The modules implemented

Netsorting allows to analyse the data using the statistical methods most used in the card sorting:

• a module for the calculus of the hierarchical cluster analysis. This modue provides the following functions:
  • computation of the proximity matrix of the items of the classification;
  • visualisation (as an html page) and export (as an excel file) of the proximity matrix;
  • computation of the hierarchical cluster analysis from the proximity matrix;
  • online visualisation (trough a web page) of the graphical dendrogram of the hierarchical analysis.
  • an interface for the manipulation of the dendrogram (status: alpha)
• a module for grouping the partecipants in different groups:
  • an algorithm to group the partecipants in different sets, based on the different conditions (for instance online versus offline, open versus closed card sorting), or random assignation; this function have been used for the experiments we are conducting using Netsorting;
  • computation of the proximity matrix for each of the partecipants groups created;
  • computation of the correlation index among the proximity matrixes of the partecipants groups, based on the algorithm used by ;
  • computation of the correlation index among the groups for each item of the cardsorting;
• a module for the analysis and the visualisation of some index of each item of the test:
  • the autocorrelation index, based on the algorithm used by ;
  • the number of times the item has been classified;
  • the list of items that have been classified in the same group, sorted by the number of common classifications;
• a web page for the visualisation of the list of partecipants of each questionniare;
• a web page for the visualisation of the data of each partecipant, based on the type of questionnaire.

The list of partecipants

A web page that links the last 100 partecipants to the questionnaire. The \hrefhttp://www.hyperlabs.net/hci/netsorting/help/dump/LikertForm.htmlexample page

The hierarchical tree

The hierarchical tree is a way to graphically show the categorisation resulted by the card sorting questionnaire. Netsorting performs the hierarchical cluster analysis from the proximity matrix. The result is showed in the page of the hierarchical tree. The \hrefhttp://www.hyperlabs.net/hci/netsorting/help/dump/analysis/hTree.htmlexample page

The item's page

How to read the results

The dendrogram can be qualitatively analysed to discover: the categories created by the partecipants; the troubleful items: items that different users have categorised in different groups. The hierarchical cluster analysis isolates those items inside the dendrogram.