Presentation Title Goes Here …presentation subtitle. Data Transformation Leilani Nora Assistant Scientist Crop Research Informatics Laboratory International Rice Research Institute :: color, composition, and layout 2 Assumptions in the ANOVA ANOVA is valid only when the following assumptions are met: • Additive Effects • Independence of errors - Assured by proper randomization • Homogeneity of variance • Normal distribution :: color, composition, and layout 3 Assumptions in the ANOVA Most common violation in the ANOVA is the homogeneity of variances. If this assumption is not met, most often the normality and additive assumptions are also violated. Transforming the data often homogenizes the variances and consequently solves the problem of non-additive effect and non-normality of variances. :: color, composition, and layout 4 Data Transformation Commonly used transformations • Logarithmic Transformation • Square-root Transformation • Arc Sine Transformation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Presentation Title Goes Here…presentation subtitle.
Data Transformation
Leilani Nora
Assistant Scientist
Crop Research Informatics Laboratory
International Rice Research Institute
:: color, composition, and layout2
Assumptions in the ANOVA
ANOVA is valid only when the following assumptions
are met:
• Additive Effects
• Independence of errors
- Assured by proper randomization
• Homogeneity of variance
• Normal distribution
:: color, composition, and layout3
Assumptions in the ANOVA
Most common violation in the ANOVA is the homogeneity of
variances. If this assumption is not met, most often the
normality and additive assumptions are also violated.
Transforming the data often homogenizes the variances and
consequently solves the problem of non-additive effect and
non-normality of variances.
:: color, composition, and layout4
Data Transformation
Commonly used transformations
• Logarithmic Transformation
• Square-root Transformation
• Arc Sine Transformation
:: color, composition, and layout5
Replace with 100-25/nFor obs = 100%
Replace with 25/nFor obs = 0%
Arcsin(sqrt(y/100))Other ranges
No transformation30 – 70
Sqrt(100-Y)80 – 100
Sqrt(Y)0 – 20
% whiteheads
% survival
Percentage data
Tiller no.
100 grain weight
Percent filled grains
Panicle number
No transformationYield and yield
components
ExampleTransformationVariable (Y)
:: color, composition, and layout6
Log(Y+1)Some obs. < 10
Log(Y)All obs. > 10
No. of insects per plot
No. of egg masses per
plant
Whole numbers that
cover a wide range
Sqrt (Y+.5)Some obs. < 10
Sqrt (Y)All obs. > 10
No. of infested plants
per plot
No. of insects caught
in traps
No. of weeds per plot
Small whole numbers
ExampleTransformationVariable (Y)
:: color, composition, and layout7 :: color, composition, and layout8