search

manybabies / mb1-analysis-public

ManyBabies1 analysis code for public sharing
MIT License
6 stars 15 forks source link

N for exclusion #2

Closed mcfrank closed 5 years ago

mcfrank commented 5 years ago

do we include labs with "more than 10 data points" (text) or with >= 10 (code)?

right now we have a conflict, not specified in https://github.com/langcog/mb1-analysis/issues/138 - I think this is me being sloppy with my > vs. >=...

thanks @angelinetsui

thoughts @christinabergmann @melsod @mekline ?

mcfrank commented 5 years ago

note that a change to the code here puts madlabucsd out of the dataset. I vote for noting the deviation from prereg and keeping them in.

melsod commented 5 years ago

I agree - honestly in colloquial speech "more than ten" is really ambiguous between the two, and I imagine whoever wrote the original text actually meant "ten or more".

christinabergmann commented 5 years ago

I also read this as "10 or more" and I did check this chunk in the code for compliance with my reading of the exclusion criteria.

christinabergmann commented 5 years ago

P.S. Later on in the preprint it says "An additional XYZ labs registered to participate but failed to collect data from at least 10 included infants, and so their data were not included. Information about all included labs is given in Table 1."

mcfrank commented 5 years ago

Noted in the deviations section.