code written: 2020-02-04
last ran: 2020-02-05
website: http://rpubs.com/navona/SPINS_outliersSD


Description. This script visualizes and summarizes outliers in the neurocognition and social cognition variables that comprise the CCA \(Y\) set. Specifically, we visualize and count outliers by 2, 3, and 4 SDs above the mean. This is in constrast to our prior analysis in which outliers were defined as 1.5 * IQR (Inter Quartile Range), where IQR is the difference between 75th and 25th quartiles.

Note. This analysis includes data from the n=410 participants that were eligible and passed DWI quality control.

Visualization. Statistical outliers are shown with larger points and a solid colour. The solid vertical line is 2 SD, dashed is 3 SD, and dotted 4 SD.

Outliers by task. The following table summarizes the data visualized above, and explicitly indicates the outlier count. Reported means and standard deviations are with outliers included. We see that there are a total of n=255 outlying values, across all social cognition and neurocognition tasks (with 214 values greater than 2 SDs but less than 3 SDs, 31 between 3 and 4 SDs, and 10 greater than or equal to 4 SDs.

mean standard deviation outlier count +/- 2 SD +/- 3 SD +/- 4 SD
neurocognition
Processing speed 45.3292683 13.4677509 15 15 0 0
Attention & vigilance 43.2059553 12.5174411 14 10 2 2
Working memory 44.5878049 11.7541377 16 16 0 0
Verbal learning 44.6219512 10.3049354 18 17 1 0
Visual learning 43.1146341 12.2613764 14 13 1 0
Problem solving 45.3390244 10.7168350 12 12 0 0
social cognition
RMET 25.8624079 4.9301263 26 21 5 0
RAD 55.6535627 8.9964624 20 20 0 0
ER_40 3391.4690594 706.4052769 20 12 4 4
TASIT_1 23.3799020 3.3073944 19 14 3 2
TASIT_2 50.4362745 7.8453014 25 19 6 0
TASIT_3 50.9262899 7.6782254 16 13 1 2
IRI 67.8487805 12.8930739 21 19 2 0
EA 0.4764206 0.1775312 19 13 6 0

Outliers by participant. Here, we see that the n=255 outlying values across all social cognition and neurocognition tasks come from n=136 unique participants. The range of outlier values per participant with an outlying value is 1: 11, with an average of 1.875. In the table below, participants without any outlying values are not shown.