Several updates (some complete, some still being debugged) but I am opening this PR asap to resolve #32 . @jessicayounger you can approve and merge this at your earliest convenience.
These release notes copy-pasted from readme:
Contains critical change to data calculations, please update ASAP.
New functions:
proc_ace_complete() is a new wrapper function that allows users who do not wish to modify any arguments to run ACE data through, start to finish. This function calls the following workhorse functions, in order. Use with care! The wrapper function effectively hard-codes argument settings for the component data processing functions. Users who wish to tinker with arguments and mid-processing choices should continue to use the functions for individual data processing steps.
load_ace_bulk()
trim_rt_trials_range(cutoff_min = 200)
proc_by_module(output = "wide")
post_clean_low_trials(.min_trials = 5)
optionally, post_clean_chance(cutoff_dprime = 0, cutoff_2choice = 0.5, cutoff_4choice = 0.25) with overall set based on user preference
post_reduce_cols() with a specific subset of output columns
and writes processed data to CSV with write_csv()
Bug fixes:
Calculation has been changed, please re-run your stuff! Thank you @mattminder for identifying that ace_dprime() was incorrectly calculating the denominator of hit and false alarm rate across all trials, instead of only across hit/miss and CR/FA trials respectively. This impacts d' metrics for SAAT and TNT data, not Filter data. (Importantly, the incorrect calculation changed d' values, tending to make them lower, but should not have affected rank-order relationships between participants' d' scores.)
post_clean_chance() was not fully updated to expect the two split SAAT modules. It should be now!
Several updates (some complete, some still being debugged) but I am opening this PR asap to resolve #32 . @jessicayounger you can approve and merge this at your earliest convenience.
These release notes copy-pasted from readme:
Contains critical change to data calculations, please update ASAP.
New functions:
proc_ace_complete()is a new wrapper function that allows users who do not wish to modify any arguments to run ACE data through, start to finish. This function calls the following workhorse functions, in order. Use with care! The wrapper function effectively hard-codes argument settings for the component data processing functions. Users who wish to tinker with arguments and mid-processing choices should continue to use the functions for individual data processing steps.load_ace_bulk()trim_rt_trials_range(cutoff_min = 200)proc_by_module(output = "wide")post_clean_low_trials(.min_trials = 5)post_clean_chance(cutoff_dprime = 0, cutoff_2choice = 0.5, cutoff_4choice = 0.25)withoverallset based on user preferencepost_reduce_cols()with a specific subset of output columnswrite_csv()Bug fixes:
ace_dprime()was incorrectly calculating the denominator of hit and false alarm rate across all trials, instead of only across hit/miss and CR/FA trials respectively. This impacts d' metrics for SAAT and TNT data, not Filter data. (Importantly, the incorrect calculation changed d' values, tending to make them lower, but should not have affected rank-order relationships between participants' d' scores.)post_clean_chance()was not fully updated to expect the two split SAAT modules. It should be now!