lucidapp

Analysis code for a project attempting to induce lucid dreams with a custom app.

One comment about notation used in all the code files. Each participant could use the app multiple times, and then could wakeup and fill out a dream report multiple times for each use. To stay consistent with the broader "participant/session/trial" language -- among other reasons -- a session refers to a single use of the app (typically a single night, but could be a nap) and a trial refers to a single dream report from within that session (typically one trial at the end of a session, but could wake up at multiple points and return to sleep).

participant - app user
session - sleep session, overnight or nap
trial - awakening and dream report

Code and file descriptions

This analysis requires the following files to exist already:

<data>/source/luciddreamdata.txt - the raw Android app output
<data>/source/reports-4ratings.xls - the experimenter dream report ratings
<data>/source/variables_legend.xlsx - experimenter-generated file that holds info about all the variables in raw output

Where the location of the <data> directory is specified in the config.json configuration file.

All data files can be found on the OSF project page.

For the <data>/source/reports-4ratings.xls file, an experimenter coded each dream report as falling into one of these categories:

lucid
semi-lucid
non-lucid
white - "I definitely had dreams, but do not remember them."
no recall - Including none, nothing to report, no dreams, I did not dream, NA, etc.
no sleep - Reported that they hadn't slept /fallen asleep yet.
not enough info - When all information is completely unrelated to dreaming/sleeping (including random characters).
not english

Non-linear files

config.json is where constants like the data directory are specified.
utils.py is where generally useful python functions are stored.

Linear files

Prepping data

# Generate the results directory structure that all files expect.
# (raw data should already be in data/source)
python setup-directories.py         #=> data/derivatives/
                                    #=> data/results/
                                    #=> data/results/hires/

# Go from raw json/txt data to csv. (It'll still be messy though.)
# Saves separate files for user data, dream report data, and app event data.
python setup-source2csv.py          #=> data/derivatives/participants.csv
                                    #=> data/derivatives/trials.csv
                                    #=> data/derivatives/events.json
                                    #=> data/derivatives/motion.json

###### ------------------------------------------------- ######
###### Manual step where someone coded the dream reports ######
###### ------------------------------------------------- ######

# Merge all the data into one file.
python setup-merge+clean.py         #=> data/derivatives/trials-clean.csv
                                    #=> data/derivatives/participants-clean.csv

Describing data

# Export some images that characterize the dataset.
python describe-samplesize.py       #=> data/results/samplesize.png
python describe-demographics.py     #=> data/results/demographics.png
python describe-correlations.py     #=> data/results/correlations.png

Analyzing data

# Test for an overall increase in LD rates with app use.
# Looks across all reports and all sessions (for those who have all 7).
python analyze-app_effect.py        #=> data/results/app_effect-data.csv
                                    #=> data/results/app_effect-descriptives.csv
                                    #=> data/results/app_effect-stats.csv
python plot-app_effect.py           #=> data/results/app_effect-plot.png

# Test if the cue had an impact on induction success.
# Looks across conditions for the first 2 sessions.
python analyze-cue_effect.py        #=> data/results/cue_effect-data.csv
                                    #=> data/results/cue_effect-descriptives.csv
                                    #=> data/results/cue_effect-stats_within.csv
                                    #=> data/results/cue_effect-stats_between.csv
python plot-cue_effect.py           #=> data/results/cue_effect-plot.png

Note you can run all this at once with runall.py

remrama / lucidapp