When we split on a the fraction of subjects (in percent, %) we want to compare the top and bottom % of subjects based on the splitting biomarker. Therefore, select % split, and the value you put in the value box underneath will set the number of subjects in each group. For example, if we put 40 into the value box, then the top group will contain the top 40% of subjects and the bottom group will contain the bottom 40% of subjects. This may be interesting if you want to compare those scoring at the extremes on a certain dimension of the ARSQ, or those having extreme values of an EEG biomarker.
The % must be less than or equal to 50% otherwise some subjects will be in both groups.
Instead of using a relative measure (% split) we can also use an absolute measure which is the value split. For example, if we want to compare the groups answering positively (ARSQ scores 4 and 5) and negatively (ARSQ scores 1 and 2), then we can use value split, and put 3 in the value box. This means the top group will contain those scoring more than 3, and the bottom group will contain those scoring less than 3. Subjects scoring exactly 3 will not be part of either group.
Once you have selected the splitting, you can see how the splitting works on the data by clicking View Splitting
In this example, we selected one group, channels, the rsq.Answers biomarker as the biomarker to split on, and we split using a value split with a splitting value of 3.
Clicking view splitting produced a number of graphs showing how the recordings would be split based on giving a score above or below three to each ARSQ item. On the x-axis of each graph we can see the ARSQ scores, and on the y-axis we can see the cumulative distribution function (CDF) which goes from 0 to 1 (0%-100% of the subjects). If we look at the first graph ( 1. I thought about my feelings), we can see a green and a red horizontal line. The green line indicates the cut-off point for the bottom group, which happens at ~0.35 (meaning the 35% of the subjects scored lower than 3 on item #1). The red-line indicates the cut-off for the top group which happens at ~0.7. All the subjects above this line scored more than 3 on the item meaning that 100-70 = 30% of subjects are in the top group.
The take-home messages are that here you have a tool to inspect how much your subjects differ in the ARSQ, and these differences we would like to find the neural correlates of in terms of EEG biomarkers.
In the following example, we split on an EEG instead of ARSQ biomarkers. We selected two groups, regions, the amplitude_8_13_Hz.Channels biomarker as the biomarker to split on. We split using a % split with a splitting value of 30 %.
Here, you can see the output for channel 62. As we selected two groups, the x-axis now shows the change in the amplitude of alpha between these two conditions. As we selected % split, the green line is set at 30% and the red line at 100-30 = 70%. Looking at where these lines cross the blue line, we can see that in the bottom group (green line) all of the subjects had a negative change in alpha amplitude of more than 0.4. For the top group (red line) all of the subjects had a very small positive change in alpha power. With NBT, we can ask whether such differences in how the alpha is affected (by the conditions) in different subjects is reflected at the level of cognition (i.e., as probed by the ARSQ).