Thanks! I enjoyed this video; however, in R, the "naniar" package's function is only reliable (and only runs) when the dataset is equal or less than 30 variables. Other functions are similarly limited. Are you aware of any alternatives?
Thanks Carla for this great question. I'm aware of the LittleMCAR function in the BaylorEdPsych package which can handle up to 50 variables. However, this package was removed from the CRAN repository. To use it, you'll need to obtain it from the archive along with the mvnmle package which is required for the LittleMCAR function, as stated here: rdrr.io/cran/misty/man/na.test.html Hope this helps. If anyone has further suggestions for Carla, please leave a comment!
Thank you for this video. I am trying to perform the Little's test using python. Most solutions online weren't really useful, so was wondering if there is a step by step methodology somewhere (video, book etc.)
If you do not reject H0 (data are MCAR), this means that there is no evidence from your observed data to suggest that the missing data mechanism is MAR. However, you cannot be absolutely certain that the data is not MAR given some variable you haven't observed. You also cannot know if the data is actually MNAR. Hope that helps.
Yes! Just what I'm studying right now 😭😭, thank you
thank you for the video. very easy to understand!
Thanks for the feedback!
I thought missing at random did not rely on itself but rather other variables?
Thanks! I enjoyed this video; however, in R, the "naniar" package's function is only reliable (and only runs) when the dataset is equal or less than 30 variables. Other functions are similarly limited. Are you aware of any alternatives?
Thanks Carla for this great question. I'm aware of the LittleMCAR function in the BaylorEdPsych package which can handle up to 50 variables.
However, this package was removed from the CRAN repository. To use it, you'll need to obtain it from the archive along with the mvnmle package which is required for the LittleMCAR function, as stated here: rdrr.io/cran/misty/man/na.test.html
Hope this helps. If anyone has further suggestions for Carla, please leave a comment!
@@statswithmia Thank you Mia!
The na.test function of the Misty package can also do it, if I'm not mistaken. :)
Thank you for this video. I am trying to perform the Little's test using python. Most solutions online weren't really useful, so was wondering if there is a step by step methodology somewhere (video, book etc.)
thanks
Using this test gives me a completely right answer: the missing data is MCAR? I really love your learning style.
If you do not reject H0 (data are MCAR), this means that there is no evidence from your observed data to suggest that the missing data mechanism is MAR. However, you cannot be absolutely certain that the data is not MAR given some variable you haven't observed. You also cannot know if the data is actually MNAR. Hope that helps.
Hi.. What if my dataset has categorical variables too? will this still work?
Yes, if the data is normal distributed (vgl. Li 2013: 797).