Actually, Pearson's kurtosis tells you nothing about "flatness" or "peakedness". It measures tails (outliers) only. Your distribution graphs actually just show differences in variance, not differences in kurtosis. You need to make the distributions have the same variance for adequate comparison. But even when you do that, it is difficult to distinguish differences in kurtosis because the tails are close to zero, even when the distribution has fat tails. (And height of peak is irrelevant, because you can have infinitely high peaks with low kurtosis and low peaks with high kurtosis). A better way to distinguish low and high kurtosis is to look at normal quantile-quantile plots, where it is easy to see the difference.
Actually, Pearson's kurtosis tells you nothing about "flatness" or "peakedness". It measures tails (outliers) only. Your distribution graphs actually just show differences in variance, not differences in kurtosis. You need to make the distributions have the same variance for adequate comparison. But even when you do that, it is difficult to distinguish differences in kurtosis because the tails are close to zero, even when the distribution has fat tails. (And height of peak is irrelevant, because you can have infinitely high peaks with low kurtosis and low peaks with high kurtosis). A better way to distinguish low and high kurtosis is to look at normal quantile-quantile plots, where it is easy to see the difference.
Thanks for explaining the interpretation of skewness and kurtosis
Can you provide the reference for this talk? I wanna quote it by reference so, Would you help?
You should share that excel file
:) how do you make the lines for the kurtosis?
The lines are just drawn. I’ll make the excel files available on our blog website
This was really helpful for my excel class but I have one question. Why aren't you outside in some party dude it's Sunday 8:33 come on !!
amen
can you provide the file for that please :(
Paypal me $10 and I will
Many thanks!!
can curtosis be exactly 0?