search

IALSA / ialsa-2015-portland-stencil

Applying lessons from the Feb 2015 Portland workshop to a newer architecture
GNU General Public License v2.0
0 stars 0 forks source link

2016-02-16 : Pre-Conference survey, domain map, REDCap #21

Open andkov opened 8 years ago

andkov commented 8 years ago

Pre-conference

the logic and logistics of the pre-conference survey

  1. email from organizers requesting to fill out the PCS.
  2. participant fill in the PCS
  3. organizers create a map of outcomes and domains that will generate the Catalog (containing MPlus scripts, customized to the participants dataset)
  4. organizers announce the domain map to the participants and request feedback (~2 weeks) to finalize the PCS.
  5. participants add more outcomes and evaluate the classification (domain map) and offer their arguments if necessary
  6. organizers process the input and finalize the PCS.
  7. The Catalog is produced produced from the finalized PCS

8 . participants start submitting individual models estimated with the the scripts provided by the catalog.

Domain map

  1. at this point the domain map is a .csv that Maleeha manually translates into REDCap logic. That's why it should be as stable as possible.
  2. after the PCS is finalized the domain map is not expected to be fluid. Once it's in the REDCap it's harder to adjust it
  3. there is an .csv map and the graph that projects this map.

    Syntax creator

  4. least wave is worse that "waves least" ; biggest concept comes first, followed by a modifier. Rename: wave_min, wave_max.
  5. XX should be MPlus illegal (e.g. %wave_min%)
  6. if waves are non-sequential, it would be easier to pass a list to Mplus. So time1-time5 should be instead time1 time2 time3 time4 time5.
  7. definitions of the timepoints and the measures in them should be dynamic (produced by the function in R, not static in Mplus).
  8. process_p_name instead of processP_name, change "process P and C" into "process A and B"
  9. Why is wave indicator must be character? Try to keep it a number.
andkov commented 8 years ago

Screenshot of the REDCap csv mapping the branching logic capture

wibeasley commented 8 years ago

The "REDCap data dictionary" storing the branching logic is stored on the repo at redcap/data-dictionaries/Ialsa2015portlandpreconference_DataDictionary.csv. The inconsistent naming scheme comes from them, not @Maleeha.

Formatted: https://github.com/IALSA/ialsa-2015-portland-stencil/blob/master/redcap/data-dictionaries/Ialsa2015portlandpreconference_DataDictionary.csv Raw: https://raw.githubusercontent.com/IALSA/ialsa-2015-portland-stencil/master/redcap/data-dictionaries/Ialsa2015portlandpreconference_DataDictionary.csv

Maleeha commented 8 years ago

@wibeasley @andkov the above two links are not working for me? I need to look at the dictionary to look at the list of possible outcomes/tests fro specific constructs under particular domains. I am looking at other stuff you referred me to.

wibeasley commented 8 years ago

It's because you change the capitalization scheme in commit https://github.com/IALSA/ialsa-2015-portland-stencil/commit/27041b836ce2dda6c3b4f54daf21e480df747941.

The data dictionary is one of the few new files where I'm ok with the wrong capitalization scheme. It's what comes out by default from REDCap, and it's probably not worth changing every time. But the date part should be removed (as it has been).