Re: st: cluster analysis: differences between clusters

[Nick Cox <njcoxstata@gmail.com>]

Interesting, but unless there is some clever trick in there I doubt
that the bootstrapping can do justice to the dependence structure of
sequence data.

Sequences are correlated (usually) and clusters are similar (by intent
if not design) and they would be much less interesting or useful
otherwise. Still an inferential quagmire and nightmare all in one...

Nick
njcoxstata@gmail.com


On 7 March 2014 15:07, Brendan Halpin <brendan.halpin@ul.ie> wrote:
> On Fri, Mar 07 2014, Andrea Jaberg wrote:
>
>> I performed optimal matching for all sequences in the dataset against
>> all others using -sqom- and the option -full-. Afterwards I grouped
>> them using cluster analysis. Now I'd like to test whether the clusters
>> are reliably different. My first thought was using ANOVA. However,
>> this seems not possible since I compared all sequences against each
>> other which results in a distance matrix.
>> What do you suggest in order to test whether the differences between
>> clusters are significant?
>
> See Studer et al "Discrepancy Analysis of State Sequences", Sociological
> Methods Research August 2011 vol. 40 no. 3 471-510
>
> This proposes a method that uses average distance to the centre of the
> pairwise distance matrix, compared with the distance to the centre of
> the partitions, to create an analogue of R-squared. It also does a
> pseudo-F test, with bootstrapped significance.
>
> It naturally fits with what you are trying to do.
>
>
> There is a reference implementation in R, written by Studer, but I have
> implemented it in Stata.
>
> I will submit this to SSC, but for the moment you can download it from
> http://teaching.sociology.ul.ie/bhalpin/discrepancy.zip
>
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/