3. Choosing Between Parametric & Non-Parametric Tests

Basic Statistical Tests
Training session with Dr Helen Brown, Senior Statistician, at The Roslin Institute, December 2015.
************************************************
These training sessions were given to staff and research students at the Roslin Institute. The material is also used for the Animal Biosciences MSc course taught at the Institute.
************************************************
*Recommended CZcams playback settings for the best viewing experience: 1080p HD
************************************************
Content:
‘Continuous’ data
-Measurements recorded on a scale
-Eg :
--White blood cell count
--Blood pressure
--Temperature
2 types of test :
--‘Parametric’ tests: suitable for normally distributed data
--‘Non-parametric’ tests: suitable for any continuous data, based on ranks of the data values
Analyses for data assumed to have a normal distribution
Checking normality
-Simple check can be based on histogram
-Satisfactory if roughly symmetrical
-Ideally data should be normal within each group tested
--In practice satisfactory if histogram for full data is symmetrical
--If full histogram has several modes, consider histograms for groups separately
Checking normality - smaller samples
-Histogram may not be smooth even if data are normal
-Difficult to determine whether normal
-Information on same measurements from previous larger studies may be helpful
-Sometimes still clear that data are non-normal, even in small sample
Data not normal?
-First consider a transformation of the data, particularly if the plots reveal a pattern
-If non-normality is due to outliers, can their deletion be justified?
-If normality in doubt :