In one of your lectures, you mention that when the amplitude is decreased, uncertainty of phase information increases. Is there a method that we can apply to decrease this uncertainty when the amplitude is decreased?
In my research, I would like to ultimately compare phase values between baseline trials and manipulated trials (amplitude reduced).
I am really enjoying your lectures on Udemy!
Hi Jason. There is no way you can decrease the uncertainty, because it’s in the measurement not the analysis. Also keep in mind that having lower power doesn’t necessarily decrease the accuracy of phase estimation; instead, lower power decreases the confidence you can have about the accuracy of the phase estimation. However, you can empirically measure the uncertainty to see whether it differs over time/conditions.
You would do this by creating empirical confidence intervals via resampling. Let’s say you have 100 trials. Select 90 trials at random and compute the phase. Then select another 90 trials at random (with replacement) and compute the phase. Keep doing that for, say, 1000 iterations, and then compute the 95% confidence intervals. You just have to be careful with the code that you are treating the values as radians, thus .1 and 3 are actually really close to each other.