Dichotomous variables are assumed to be non-normally distributed. When testing for normality using Skewness and Kurtosis, the values I get for the dichotomous variables are within the accepted range.*
Am I doing the right test or am I really supposed to get values out of the range?
*Skewness: absolute values under 2
Kurtosis: absolute values under 4
Thanks
In my opinion, it simply does not make sense to examine skewness and kurtosis for dichotomous variables. The values that you get are meaningless, regardless of whether they "pass" some arbitrary threshold.
The quesstion I have: for what do you need to check for normality in a dichotomous variable?
Just for the sake of clarity and because this comes up here quite frequently: the normality assumption in the linear model (e.g. t-test, ANOVA, regression) is neither about the distribution of the predictors, nor the "data" itself, i.e. the dependent variable. Therefore, if your dichotomous variable will be your predictor in any linear model, there is no need to check for normality, since it is not an assumption.
Thanks for the answer Christian Geiser .
Rainer Duesing thank you again for explaining. I understand that assumption is about the dependent variable and the normality of the independent is not needed in a report. The purpose of the question was to try and understand why I am getting the values within the range in the mentioned tests. Look like it's not important to understand or needed.
Thanks.
In addition to Christian Geiser 's point about it not making sense to use these with binary variables, it is worth adding skewness and kurtosis tests are not tests of normality so not sensitive to lots of deviations.
As colleagues have already pointed out, for a dichotomous random variable (a discrete random variable with two values), it makes no sense to investigate asymmetry and kurtosis. Write down exactly what you want to investigate for your dichotomous random variable and, perhaps, there will be an answer in another plane.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation