Check Figure 2

[KirstensGitHub]

At today's meeting we discussed checking the generation of each stat and figure from the beginning (data compilation) to the final stat/figure.

This issue will be a review of the checks done for FIgure 2 - violin plots

[KirstensGitHub]

INITIAL CHECKS:

• behavioral data check (see issue #77)

• review and comment function for generating all ttest lines

• review and comment code for ttest stars

• review code that adds all lines to violin plot in descending order

FINAL CHECK:

• plot the data and all stats tests and confirm that plot visualizations match test outputs

[KirstensGitHub]

Below is the hand check for the bars in the violin plot for Experiment 1. For each significant relationship, there is a letter to the right of each t-test indicating the letter/label of the line in the plot representing the relationship.

Below is the hand check for the bars in the violin plot for Experiment 2. For each significant relationship, there is a letter to the right of each t-test indicating the letter/label of the line in the plot representing the relationship.

Below is the hand check for the stars in the violin plot for Experiment 1.

Below is the hand check for the stars in the violin plot for Experiment 2.

[KirstensGitHub]

@jeremymanning this issue is ready for review !

[jeremymanning]

It looks like this is a test of the plots (looks sufficient). We should also test that the numbers that go into the plots are correct. E.g. spot check a few examples for each stimulus and attention category to make sure the within-participant means are correct, and then also check that the across-participant means are correctly reflected in the violin plots.

[KirstensGitHub]

@jeremymanning that sounds good! I use the pandas groupby functions to group by subject, attention level, and image category (with the image labeling having already been checked in issue 77) and take the mean, so I think this check would be the equivalent of checking the pandas groupby function. (The way pandas averages over nans is what tripped me up with the small sliding window difference, so it's not unreasonable to check, just making sure we're on the same page.)

[KirstensGitHub]

info on pandas groupby:

https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.groupby.html