Contains experiment and analysis scripts, alongside data files and analysis
outputs.
## Repository Structure ##
### Top-level directories ###
- **pilot** : Materials for pilot experiment. Employs same design as
Experiment 1, except only two adaptation durations (35 and 140 s) are
included.
- **experiment1** : Materials for first experiment, testing effects of
adaptation duration and visual spatial references frames on the
ventriloquism aftereffect.
- **experiment2** : Materials for second experiment, testing effects of
constant versus variable audio-visual disparities on the ventriloquism
aftereffect.
### Sub-directories ###
All top-level directories follow a similar sub-directory structure:
- **analysis** : Contains analysis scripts (see below), and outputs of those
analyses. Contains additional sub-directories:
- ***ind*** : Outputs of first-level simple regression analyses. Outputs
for each subject are within further separate sub-directories. Actually
uploaded as a zip file due to the large number of sub-directories.
- ***group*** : Outputs of second-level analyses (ANOVAs, t-tests, etc.) on
the first-level spatial bias and gain parameter estimates.
- ***mixed_effects*** : Outputs of alternative analysis pipeline, in which
the first-level analysis instead utilises mixed-effects regression
models. Sub-directories *level1* and *level2* correspond to the first-
and second-level analysis outputs respectively (equivalent to the *ind*
and *group* outputs). Note that this pipeline has been deprecated in
favour of the simple regression approach due to the mixed-effects
parameter estimates being biased.
- ***per_cycle_analysis*** : Added in response to reviewer comments.
Provides equivalents of first- and second-level VAE outputs but for each
adapt/test cycle separately.
- **data** : Contains raw data log files.
- **experiment** : Contains materials and code for running experiment.
## Analysis Scripts ##
All experiments employ the same basic analysis scripts:
- **statsUtils.R** : Contains generic utility functions for use by other
scripts.
- **doIndRegressions.R** : Runs first-level simple regression analyses for
each subject independently. Extracts spatial bias and gain estimates for
each subject.
- **doMixedEffRegressions.R** : Runs alternative first-level analysis
pipeline, instead employing mixed-effects regression models with the
subjects entered as a random-effects factor. This has been deprecated in
favour of the per-subject simple regression approach as the mixed-effects
parameter estimates are liable to be biased towards the mean (making them
sub-optimal as inputs to the second-level analyses).
- **doGroupAnalysis.R** : Performs second-level analyses (e.g. ANOVAs,
t-tests, etc.) on the spatial bias and gain estimates produced by the
first-level analysis. Accepts inputs from either the simple or mixed-effects
regression pipelines (can control which with a command-line flag; defaults
to the simple regression pipeline).
- **doPerCycleAnalysis.R** : Added in response to reviewer comments. Performs
first- and second-level analyses of VAEs for each adapt/test cycle
separately.
