search

IALSA / ialsa-2016-time-sensitivity

Investigating sensitivity of statistical conclusions to features of longitudinal design
GNU General Public License v3.0
0 stars 0 forks source link

MMSE plots #1

Open andkov opened 7 years ago

andkov commented 7 years ago

@GracielaMuniz @ampiccinin @annierobi

Linear models

Quadratic models

ampiccinin commented 7 years ago

@andkov @GracielaMuniz @annierobi

Odd that the quadratic don't seem much less linear, but that the estimates go above 30, where they don't with the linear.

Were these done in Mplus? Was tobit used? For quadratic, but not linear, perhaps?

Nice graphs!

andkov commented 7 years ago

Yes, all these were done in Mplus and processed through Portland-like pipeline. I didn't upload the outputs because they were too many,but in retrospect it appears an error of judgement. I'll push the Mplus files underlying these graphs later today when I get to my laptop - it's no trouble.

annierobi commented 7 years ago

Tobit was not used for these models. Unless new models were run that I am unaware of. The models in dropbox do not use tobit.

On Tue, Jan 17, 2017 at 5:22 PM, Andriy V. Koval notifications@github.com wrote:

Yes, all these were done in Mplus and processed through Portland-like pipeline. I didn't upload the outputs because they were too many,but in retrospect it appears an error of judgement. I'll push the Mplus files underlying these graphs later today when I get to my laptop - it's no trouble.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/IALSA/ialsa-2016-time-sensitivity/issues/1#issuecomment-273319485, or mute the thread https://github.com/notifications/unsubscribe-auth/AKow3dMZuJoCh8sq1_Z1ZtHfTaUJ89gcks5rTT8QgaJpZM4LiLYW .

andkov commented 7 years ago

The models on Dropbox that I used as guides for automation did not use Tobit. The models behinds these graphs also don't use it.

Sent from my iPhone, please excuse typing and T9 errors.

On Jan 17, 2017, at 12:37 PM, annierobi notifications@github.com wrote:

Tobit was not used for these models. Unless new models were run that I am unaware of. The models in dropbox do not use tobit.

On Tue, Jan 17, 2017 at 5:22 PM, Andriy V. Koval notifications@github.com wrote:

Yes, all these were done in Mplus and processed through Portland-like pipeline. I didn't upload the outputs because they were too many,but in retrospect it appears an error of judgement. I'll push the Mplus files underlying these graphs later today when I get to my laptop - it's no trouble.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/IALSA/ialsa-2016-time-sensitivity/issues/1#issuecomment-273319485, or mute the thread https://github.com/notifications/unsubscribe-auth/AKow3dMZuJoCh8sq1_Z1ZtHfTaUJ89gcks5rTT8QgaJpZM4LiLYW .

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

andkov commented 7 years ago

@ampiccinin The output files have been uploaded to https://github.com/IALSA/ialsa-2016-time-sensitivity/tree/master/output/studies/octo/