Hi. I just joined statalist and ask the first question as a beginer.
I'm worrying about duplicate question, but ask for individualized question. Please give me a thoughtful understanding.
I'm on analysis for risk factor of development of proteinuria.
Sample size is 319,457.
To list and compare variables according to some categories (BMI categories or Presence of proteinuria) ,
I need to know whether the variables follows a normal distribution or not.
(Because to use independent t-test or ANOVA for normal distribution continuous variables, and Kruskal-Wallis test for non-normal distribution continuous variables.)
So I performed sktest for normality test.
Skewness/Kurtosis tests for Normality
------ joint ------
Variable | Obs Pr(Skewness) Pr(Kurtosis) adj chi2(2) Prob>chi2
-------------+---------------------------------------------------------------
wbc | 317,927 0.0000 0.0000 . .
bmi | 317,927 0.0000 0.0000 . .
sbp | 317,927 0.0000 0.0000 . .
height | 317,927 0.0000 0.0000 . 0.0000
Q1) Is it right to interpret this result as not following the normal distribution?
Q2) Why wbc, bmi, sbp doesn't report Prob >chi2 , but height report 0.000
When I performed sum, detail for height , it's skewness was 0.05 and kurtosis 3.21
And.. when I draw histogram, like below.
Array
Q3) Not strictly, can I judge this variable as a normal distribution variable ???
I think the sample size is too large so sktest reports does not follow the normal distribution.
Q4) Is there any appropriate test for large number of sample data??
And. Last...
I learned Central Limit Theorem (when sample size is large enough, we can assume the data follows normal-distribution.)
Q5) Can I apply this theorem to my analysis?? It means, whether I can use mean value instead of median value and use ANOVA test instead of Kruskal-Wallis test .
Related Posts with Normality test*for large sample data
Transfering Kendal correlations into WordDear all contributors, I am calculating Kendal correlations of more than 100 variable couples and w…
Consistent estimatorDear all, Suppose we have a lot of observations. Suppose the following model : y = b*x + e , where …
Fixed Effects Poisson or Negative Binomial?Hi, I am trying to choose between a Fixed-Effects Poisson regression model and a Fixed-Effects Nega…
Generating new columns for each unique value of a categorical variableDear All, The issue I am trying to resolve here is as follows- I have a dataset where customers pur…
gender changes in the panel surveyDear Statalist Community, due to my thesis I am allowed to work with stata. My dataset comes from a…
Subscribe to:
Post Comments (Atom)
0 Response to Normality test*for large sample data
Post a Comment