In horticulture science, we generally studied variability in germplasm. We follow correlation analysis to relate all variable under studies. We generally do Pearson correlation. But it is not always suited to horticultural or related data. Present of extreme value is much common phenomenon which adversely affect the normality of the data. It is particularly happen when we select wild relative allong with common cultivated variety. Sometimes it also observed for land races. It also happened for microbiological data of fresh and processes products. Sometimes it happens to data related to the effect of adverse climatic condition on plant performance. In wetlab chemistry it sometimes tapped to different solvents.
So all these condition proper correlation studies are essential. Selection of suitable techniques of correlation helps to interpret the data more acirately.
Recently I gain knowledge in these field. If you feel interest kindly respond me.
What are the ideal conditions where we should prefer multiple correlation over PCA.. any guess..??
DATA CLASSIFICATION: review your data classification for Y (dependent variable) and X (independent variable). Each may be classified to be anyone of these; (i) quantitative, (ii) ordinal, and (iii) nominal. The type of correlation coefficient used depends on the data type of Y and X. Here are possible combinations and the corresponding correlation coefficient.
XqYq = Pearson r
XqYo = Biserial
XqYn = Point biserial
XoYq = Biserial
XoYo = Spearman rho, Tetrachoric & polychoric
XoYn = Rank biserial
XnYq = point biserial
XnYo = Rank biserial
XnYn = Phi, L, C or lambda
... where X = independent variable; Y = dependent variable; q = quantitative; o = ordinal; and n = nominal data.
Each have their own assumptions. for example. Pearson r Each observation independent from each other. continious variables approximetaly normally distrubuted, a linear relationship exist between variables.Borrowed from an answer
- Badges
- Science topic
- Similar topics
- Biological Science
- Microbiology