Transforming a negatively skewed ikert scale dependent variable for ANOVA and M.Reg
- Thread starter Muppet
- Start date
- Tags likert negative skew transformation
Re: Transforming a negatively skewed ikert scale dependent variable for ANOVA and M.R
why do you think that you want transform your dependent variable? Linear Regression or ANOVA don't assume normally distributed dependent variables (the residuals should be sampled from a normally distributed population, and even this is assumed only for smaller samples).
With kind regards
K.
why do you think that you want transform your dependent variable? Linear Regression or ANOVA don't assume normally distributed dependent variables (the residuals should be sampled from a normally distributed population, and even this is assumed only for smaller samples).
With kind regards
K.
Re: Transforming a negatively skewed ikert scale dependent variable for ANOVA and M.R
Hello K. Thanks for the reply. I was thinking that the skew would affect the distribution of the residuals(and does seem to have in a couple of places). It's annoying as I am examining stigma so it is generally going to be skewed data. I am now thinking of using bootstrapping when I run the tests.
Hello K. Thanks for the reply. I was thinking that the skew would affect the distribution of the residuals(and does seem to have in a couple of places). It's annoying as I am examining stigma so it is generally going to be skewed data. I am now thinking of using bootstrapping when I run the tests.
Re: Transforming a negatively skewed ikert scale dependent variable for ANOVA and M.R
Hello K., my samples are 78, 79 and 80 - not so small. I am a bit of a stats muppet but I think you are saying that even with stigma Likert scale measures if the sample is large enough I don’t need to worry about normality thanks to the central limits theorem? But if that is the case why is there ever an issue deciding between parametic vs. non parametric measures in the case of non normality in samples above 30? It’s all a bit confusing for me…
Hello K., my samples are 78, 79 and 80 - not so small. I am a bit of a stats muppet but I think you are saying that even with stigma Likert scale measures if the sample is large enough I don’t need to worry about normality thanks to the central limits theorem? But if that is the case why is there ever an issue deciding between parametic vs. non parametric measures in the case of non normality in samples above 30? It’s all a bit confusing for me…