TODO 16.11.18 after rerunning data with updated functions

[k47ha]

• [x] make history labels / commits on overleaf when adding (for comparison)
• [x] make new figures
• [x] make small grid statistics
• [x] program microsaccades again and try lambda of 5,8,15 (put it into the notebook)
• [x] put results about saccade amplitudes into the text
• [x] rewrite discussion
• [x] find replace * to the \SI
• [x] Head movement: Labeling colors is missing for the different eye trackers. No explanation for different shades of grey.
• [x] be able to use PeerJ format
• [x] send draft to peter in PeerJ format on Wednesday 21.11
• [x] write test for 3d angle (accuracy)
• [x] put doi
• [x] rerun small grid statistics
• [x] remove whitespace around figures
• [x] renew screenshot (how do I make a screenshot in a high quality?)
• [x] renew black and grey offset plot from VP25 block2 with Helvetica font
• [x] put new smooth pursuit and freeview saccade amplitudes into the text
• [x] renew good raw signal after preprocessing with Helvetica font
• [x] adjust style of Figure 5 also content
• [x] Figure 9 (Results Freeviewing) is misssing which color is which eye tracker. Or we could mention that in caption?
• [x] put dondors in affiliation?
• [x] make data available at OSF (we also promised gif of grid task)
• [x] send Peter a Mail But often, worse performance can be compensated by more trials or more subjects. But such "noise suppression" helps only with the square root of N. A back-on-the-envelope calculation is: To compensate for the 40\% worse accuracy, one would need to record twice as many subjects (sqrt(2) $\approx$ 1.4) with Pupil Labs than with the EyeLink~1000 \citep[compare also to ]{wilming_measures_2011}. change font in theme possible eye tracking test battery