Underlying stats for MaxQDA code maps

[etopp]

Hello,

I would like to know more about the MDS used for the "Code Map" function in MaxQDA Visual Tools. In the MaxQDA Manual, it states that the Code Map is "based on classical MDS". This process requires generating a dissimilarity matrix from the Codes Relations Browser, and then converting that into a distance matrix.

Can you explain more about what type of method is used in this ordination, for example PCoA or NMDS? And which method is used to create the distance matrix, e.g. is it Bray-Curtis or Gower's distance? Also, is it possible to generate stressplots or some kind of measure of goodness of fit for these ordinations?

This would really help, as I am trying to understand why I find differences in mapped codes between the Code Maps and other ordinations I have made with my data.

Thank you!

Version: MAXQDA 2020
System: Windows 10

[Maren_eng]

Hello etopp,

Thank you for your post! Let us go through your questions one by one:

1. Calculation of the distance matrix from the similarity matrix
As the manual states, the conversion procedure is as follows:
"For this purpose, a similarity matrix is first calculated according to the display in the Code Relations Browser and then converted into a distance matrix. For the conversion, the column sums are calculated first, so each code is checked to see how often it occurs together with any other code. The maximum of these column sums is determined and defined as the maximum possible similarity. In each cell, the similarity of two codes is subtracted from this maximum." (https://www.maxqda.com/help-mx20/visual-tools/code-map-position-codes-according-to-similarity)

2. Classical Multidimensional Scaling
MAXQDA uses the same algorithm as presented in Wikipedia, so it should correspond to PCoA:
https://en.wikipedia.org/wiki/Multidimensional_scaling#Classical_multidimensional_scaling

3. Stress test or Quality Measure
Unfortunately MAXQDA does not offer these values

Kind regards,
Maren