search

langcog / metalab2

MetaLab -- Community-augmented meta-analysis
http://metalab.stanford.edu/
MIT License
21 stars 8 forks source link

new version of vowels MA with talker normalization information #65

Open alecristia opened 6 years ago

alecristia commented 6 years ago

I have prepared a new version using a script that splits information formerly bunched into one column: https://docs.google.com/spreadsheets/d/1n3BQj8fimL8LDJ72Y06zBpj_ysSxolCP1dwho7JsER8/edit#gid=1372713622

There are some apparent divergences because R removes field-initial spaces or field-final spaces (so there are differences between original " frequency counts" and new "frequency counts"). There are also some divergences for which I don't see a difference: 0.5869565217 versus 0.5869565217

But overall, there are no bugs and it seems my script correctly added back the columns that had been merged.

Could I have a second pair of eyes look through before we replace the current vowels?

Thanks in advance!

alecristia commented 6 years ago

Addition: In the "new" version I've used black background + white font to mark the columns that I'm NOT going to add back (because they didn't work well in the split script, probably because the contents were human-readable only; notice factors and numeric columns imported okay).

alecristia commented 6 years ago

Red background = new columns I'll be coding

alecristia commented 6 years ago

I'm marking errors I'm finding in the original tab as comments

shotsuji commented 6 years ago

Yes will do within a week! Thanks for updating!

cissard commented 5 years ago

HI Sho,

Can I close this issue or do you want me to look at it?

Cécile

shotsuji commented 5 years ago

Oh oops I never did that! If you want to have a look great - lmk if you need more info or help!

cissard commented 5 years ago

Hi,

I corrected the mismatchs that were due to field-initial or field-final spaces. However I left the long comments unchanged to let you decide what makes sense from an expert of the field point of view. Also about the corrupted text it's in both old and new versions. These are only comments so it shouldn't affect the MA results and compuatitons.

About the numeric divergences the one that I found came from the fact that R took the rounded number displayed by google sheet, not the exact number inputed in the cell formula, so I copied the exact umber in the new version.

@alecristia I let you close the issue if it's good for you.