My data transformation not working (despite You Tube showing other peoples data which transforms beautifully). I need to transform the data as it is my main outcome variable and I need to conduct parametric (e.g. regression) analyses.
I am trying to transform positively skewed scores on a scale (1-5) with Log10 but am still not able to reach normality ( i.e., skewness within a Z-score range of +/-1.96; Kurtsis is good enough).
Could it be because 48% of my data is on a single value? (ie, 1; scale is 1-5. 1 indicates 'no problem', so 2-5 are mild-extreme problem). I have tried many other transformations ( LN, Sqrt etc) and have consulted Pallant and Tabachnick & Fidell but to no avail.
I have attached the histogram and descriptives table here.
All suggestions appreciated.
By having a single value repeated by 48% of sample size, it becomes obvious that your data is not continuous. So that, none of these approaches above would help you. But, you may decide to "divide" your data into two pieces; the "single value" one and the continuous one. Later then you may provide some outcomes for continuous one.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics