Plan for Session 3: Annotation and Analysis of Data II
(continuation of Session 1)


FIRST PART: fix XML files and finish annotation
---------------------------------------
- fix the wrong XML files. 
  - make sure that the fixed files can be opened
- finish annotation
- those that have finished, go over the annotation with the "search" tool to spot possible mistakes

SECOND PART :  calculate inter-annotator agreement with kappa
------------------------------------------------------
  - pairs that have annotated the same file, compress the directory and send it by email to your partners (and to me)
  - short intro to kappa statistics
  - how to use the statistics facility of MMAX
  - calculate kappa:
    	      - speaker role (it should be 1, perfect agreement)
	      - marker (idem)
	      - dysfluency type
	      - forward looking
	      - backward looking
   - analysis of results: 
     	      - where did you encounter disagreements?
	      - hypothesis of why?
 
- come forward in pairs and present results.

THIRD PART: statistical comparison of conditions with t-test
--------------------------------------------------
- short intro to t-test
- how to use the online t-test calculator
- each of the two groups has 6 files with 3 different conditions that can be compared: 
       - Turn taking (3 FTT/3 P2T)
       - Speaker role (6 players / 6 executors)
       - Gender (3 female players/ 3 male players)

- possible measures (open list - creativity welcome!)
  - average of dysfluencies per turn across conditions
  - number of forward looking acts across conditions
  - number of backward looking acts across conditions
  - ...

- make hypothesis to try to explain the results obtained (especially when 
there is an statistically significant difference across conditions)

- come forward in groups and explain results.