Eye-eye lag analysis

To perform eye-eye lag analysis, you will need to run the cross-correlation.py analysis. This takes a GDF file (and optionally a list of times) as input and outputs a csv file where each row represents a time during the experiment (by default, the midpoint of each trial). The initial columns record the time, Trial ID and number of current parts, and the remaining columns interleave the eye-eye lag data at the specified intervals with a randomised baseline taken by randomising the existing eye data across time for each participant.

For further analysis, you may want to use the reformatcc.py script which takes the output of this analysis and a file containing experimental conditions for each trial and produces 8 csv files, one for each experimental condition. If cross-correlation.py is run with multiple file names, reformatcc.py with gather conditions together across experiments.

Usage: [python] cross-correlation.py [options] filename.gdf
Options:
	-s n, --samplerate=n	length of eyetrace sample (ms)
	-m n, --maxlag=n	maximum lag magnitude (ms)
	-l n, --lagstep=n	lag step size (ms)
	-w n, --width=n		size of window (ms)
	--spread=n		see width
	-p, --partsonly		don't count looks on regions as hits
	-i n, --inputfile=n	File containing list of times (- for stdin)
	-b n, --boss=n		agent/lagtype for primary timing information
	-e n, --employee=n	agent/lagtype for secondary timing information

For boss and employee, agent/lagtype defines the events over which lag is calculated. The first part (agent) matches a subject id (e.g. client) and the second part (lagtype) matches a timed event (usually look or hover). In all calculations, the lag is defined from the boss to the employee, so a positive lag indicates that the employee event is after the equivalent boss event, and a negative lag indicates that the employee event comes before the equivalent boss event. If one is defined, both must be defined. If neither are defined, boss = a/look and employee = b/look.