Hey man! I'm not a 100% percent sure, I've only been fooling around with the data. It seems that kstest compares the distribution 'x' to a standard normal distribution. Meaning that if your data is not centered on 0, you're already going to be set for a low p value. If you fit a normal distro with the sample mean and std you get results more frequently concordant with Shapiro Wilks.
Hi! Thank you for the video it's very clear & easy to follow! I have data from an experiment in the form of accuracy (responded correctly =1 & incorrectly =0) across various participants. How would I test if this data has a normal distribution?
It would depend on what features you collected. It might be difficult to rely solely on the 0/1 from a single question. If you have multiple 0/1's you could try using something like a summation for their overall scores.
Hey man! I'm not a 100% percent sure, I've only been fooling around with the data. It seems that kstest compares the distribution 'x' to a standard normal distribution. Meaning that if your data is not centered on 0, you're already going to be set for a low p value. If you fit a normal distro with the sample mean and std you get results more frequently concordant with Shapiro Wilks.
Suggestion - Please use a rather simple continuous variable.
Good video.
Thank you so much Adam.
Hi! Thank you for the video it's very clear & easy to follow! I have data from an experiment in the form of accuracy (responded correctly =1 & incorrectly =0) across various participants. How would I test if this data has a normal distribution?
It would depend on what features you collected. It might be difficult to rely solely on the 0/1 from a single question. If you have multiple 0/1's you could try using something like a summation for their overall scores.
@@AMGaweda Thank you! Also when testing for normality, does one do this across all participants or within each participant?
Tas cabron, gracias we!