If the SDs look particularly small or large (e.g., <10% or >35% of the range for Likert-type data), consider using SPRITE to see what the data might have looked like to produce those SDs. Paper: https://t.co/Rgf1PsH12S Shiny app: https://t.co/BRTOfHBYMr /10
Maybe it's worth thinking about whether something along the lines of @jamesheathers & @sTeamTraen's work in psych can be applied here (although I think that mostly works on integer data)
https://t.co/0cnbf4OSIl
GRIM test: https://t.co/ZPSLtHPcrV
SPRITE: https://t.co/rAJnWtXBl5
If you wish to host #RShiny apps on Code Ocean, this capsule provides a template and Readme for running one: https://t.co/M0iFX9JTBI
The underlying Rsprite code is courtesy of @sTeamTraen, which you can read more about here: https://t.co/DKXUl16rRO
@wolfvanpaemel Good question. Yeah there's a formula for the 'full canopy' i.e. the line created using sets of (in this case) 0 and 400. This sort of error corresponds to the green box in Figure 1B.
https://t.co/WO3JaJJtEt
@WrightingApril @MarkScherz @allopatry @Vert_Anatomist @frogsicles @frogdiaz @TrackingActions @ProfRomi @DanielBolnick I haven’t used it myself but that’s exactly the idea behind SPRITE.
https://t.co/uaDJBZqUQm
@chanelkmeyers @nhcharlot Tools like SPRITE and GRIM seem like they are helpful for these tasks @jamesheathers @sTeamTraen Have not used them myself yet, but see https://t.co/ckgAjyUh3K
@Rob_Tarzwell @barttels2 @stephensenn @ChristosArgyrop @VPrasadMDMPH @anish_koka @venkmurthy @adamcifu @pash22 @Portland_State @peterboghossian @oncology_bg @FearLoathingBTX @ArthurCaplan @AlexJohnLondon That's why we double down on ways to point that out simply.
https://t.co/qnamTj8d3T
https://t.co/Rhhk7lcjsN
@causalinf @KiraboJackson Publishers should automate some of that checking, so reviewers can focus on the other stuff.
https://t.co/3vCx2mOPLI
https://t.co/dzxYb94YJV