Thank you for the excellent video! One question: Shouldn't the values you mentioned for skewness and kurtosis be the z scores? I've read elsewhere that the values given in the table for e.g. skewness needs to be divided by their Standard Error value to obtain the z-score, which is the score we're interested in, not the value shown in the table. Thanks!
Appreciate your work much. Thank you. The more accessible manner ever we see to manage and report and interpreted our data is this software. Is amazing supper software. My question is, if the data deviates typically from the normal, how can we transform and correct our data. Thank you and hope you'll reply to my inquiry.
You can do a log-transformation; alternatively, identify the outliers through using the boxplot (see my relevant videos) OR calculate the Mahalanobis's distance (see my video on regression) and remove the outliers.
The test statistic (W) measures how closely the data follow a normal distribution. A value of W close to 1 indicates that the data are normally distributed, while a value close to 0 indicates a departure from normality. The p-value (P) is the probability of obtaining a test statistic as extreme as the one calculated, assuming that the null hypothesis is true. Interpret this like usual p-values. If the p-value is less than a chosen level of significance (usually 0.05), then we reject the null hypothesis and conclude that the data do not come from a normal distribution.
Very helpful video! Can I ask what was the name of the textbook you referenced as far as interpreting skewness is concerned? I am dealing with a multilevel model and I was trying to find sources on how to interpret the skewness of my residuals.
Please see this book/ chapter 1: www.routledge.com/Quantitative-Data-Analysis-for-Language-Assessment-Volume-I-Fundamental/Aryadoust-Raquel/p/book/9780367671396
@@VahidAryadoust Thank you a lot! Unfortunately I am not anywhere near language assessment as far as my project is concerned, so I think buying a book specific for that would not be efficient, I am sure you will understand. Can I ask if you have any other references for interpreting the absolute number of skewness and saying that the shapiro wilk significance testing is not very reliable?
@@VahidAryadoust Thank you very much! I wasn't able to find anything on the shapiro w test being unreliable for low samples (my problem is that for shapiro w test it's a non normal distribution but by using the absolute value for kurtosis and skewness it is normal so I don't know what to use), but I did find the rule of thumbs as far as skewness and kurtosis are concerned, I think I will simply follow that rule, thank you a lot for your help and the videos, I have already recommended them to a friend of mine, they are really excellent!
How did you calculate the composite scores of the questionnaires and test scores as presented in your paper? did you take the sum of the items or the mean of the items included in each domain/section?
Hello Dr. Aryadouust. Can Likert scale data be treated as continuous? the skewness and kurtosis of my DVs fall between normal and acceptable ranges, i.e., ±2 for skewness and kurtosis. Sample size is 124. However, all DVs are ordinal (5 and 6-point Likert scales). Do you suggest running repeated measures ANOVA to measure the effect of intervention on two groups over two times, or shall stick to the non-parametric test (Wilcoxon singed-rank test). Thank you in advance.
It is common in Social sciences to treat them as continuous data, but personally I prefer to treat them as categorical, unless you aggregate a few items to create a 'super-item'.
@@VahidAryadoust okay in the case of Kurtosis, some say that it should be btw -1 and +1 to show normality but others say that it should be btw -2 and +2 ... which coefficiant should I consider??
I just got a doubt. What if the predicting regression model produces skewness, can the performance of the model be judged looking at the skewness (either left or right)? What inferences can we draw?
![Normality test [Simply Explained]](http://i.ytimg.com/vi/AVketBmpUTE/mqdefault.jpg)
![Normality test [Simply Explained]](/img/tr.png)
Thank you for the excellent video!
One question: Shouldn't the values you mentioned for skewness and kurtosis be the z scores? I've read elsewhere that the values given in the table for e.g. skewness needs to be divided by their Standard Error value to obtain the z-score, which is the score we're interested in, not the value shown in the table.
Thanks!
That idea also works. I have another video explaining z scores for skewness and kurtosis if you can browse through the playlist.
@@VahidAryadoust Thank you!
Appreciate your work much. Thank you. The more accessible manner ever we see to manage and report and interpreted our data is this software. Is amazing supper software. My question is, if the data deviates typically from the normal, how can we transform and correct our data. Thank you and hope you'll reply to my inquiry.
You can do a log-transformation; alternatively, identify the outliers through using the boxplot (see my relevant videos) OR calculate the Mahalanobis's distance (see my video on regression) and remove the outliers.
Very nice video!
wow great lecture sir❤
Hi ! Please what's the difference between ShapiroWilk W and ShapiroWilk P ?
The test statistic (W) measures how closely the data follow a normal distribution. A value of W close to 1 indicates that the data are normally distributed, while a value close to 0 indicates a departure from normality.
The p-value (P) is the probability of obtaining a test statistic as extreme as the one calculated, assuming that the null hypothesis is true. Interpret this like usual p-values. If the p-value is less than a chosen level of significance (usually 0.05), then we reject the null hypothesis and conclude that the data do not come from a normal distribution.
with the total sample 405, shapiro wilk (p value) is false?
Very helpful video! Can I ask what was the name of the textbook you referenced as far as interpreting skewness is concerned? I am dealing with a multilevel model and I was trying to find sources on how to interpret the skewness of my residuals.
(I also wanted to read your paper but sadly my institution does not provide access to this specific paper for some reason).
Please see this book/ chapter 1:
www.routledge.com/Quantitative-Data-Analysis-for-Language-Assessment-Volume-I-Fundamental/Aryadoust-Raquel/p/book/9780367671396
@@VahidAryadoust Thank you a lot! Unfortunately I am not anywhere near language assessment as far as my project is concerned, so I think buying a book specific for that would not be efficient, I am sure you will understand. Can I ask if you have any other references for interpreting the absolute number of skewness and saying that the shapiro wilk significance testing is not very reliable?
@@JOYLESSSSS i understand; you can find useful references in the most recent videos I have just released: ruclips.net/video/Kw5zaxB8Zac/видео.html
@@VahidAryadoust Thank you very much! I wasn't able to find anything on the shapiro w test being unreliable for low samples (my problem is that for shapiro w test it's a non normal distribution but by using the absolute value for kurtosis and skewness it is normal so I don't know what to use), but I did find the rule of thumbs as far as skewness and kurtosis are concerned, I think I will simply follow that rule, thank you a lot for your help and the videos, I have already recommended them to a friend of mine, they are really excellent!
How did you calculate the composite scores of the questionnaires and test scores as presented in your paper? did you take the sum of the items or the mean of the items included in each domain/section?
Yes, sum them up to create aggregate scores. Alternatively, you can compute the mean; either will be fine, but I prefer the mean-based method.
@@VahidAryadoust Clear. Thank you so much.
Hello Dr. Aryadouust. Can Likert scale data be treated as continuous? the skewness and kurtosis of my DVs fall between normal and acceptable ranges, i.e., ±2 for skewness and kurtosis. Sample size is 124. However, all DVs are ordinal (5 and 6-point Likert scales). Do you suggest running repeated measures ANOVA to measure the effect of intervention on two groups over two times, or shall stick to the non-parametric test (Wilcoxon singed-rank test). Thank you in advance.
It is common in Social sciences to treat them as continuous data, but personally I prefer to treat them as categorical, unless you aggregate a few items to create a 'super-item'.
Excelent!
@Statistics & Theory Nice video👍I wish to know if I want to show normality should I check skewness and Kurtosis or use Shapiro Wilk test ?
Either will do. I would look into skewness and kurtosis.
@@VahidAryadoust okay in the case of Kurtosis, some say that it should be btw -1 and +1 to show normality but others say that it should be btw -2 and +2 ... which coefficiant should I consider??
I just got a doubt. What if the predicting regression model produces skewness, can the performance of the model be judged looking at the skewness (either left or right)? What inferences can we draw?
Please watch this for this question:
ruclips.net/video/3sQnO02f8Z0/видео.html
+/- 2 for skewness
+/-7 for kurtosis
... compatible with normal distribution
world's driest throat