HI!
I am wondering what does the spearman 's correlation tells in biological terms?
So, assuming you have equal number of samples per condition A and B and the rho= 1 this would mean a perfect positive correlation but in biological terms what does it me?
spearman's does not assume that what happen in A would have an effect in B (that's pearson's if I am not wrong) but so how do you interpret that coefficient?
thanks
camilla
Simply that there is a perfect association, so that a person's rank on one variable can be used to predict precisely what that person's rank on another would be. While correlations can never be used to establish causality, a correlation that strong suggests either the two variables are causally related or very tightly related to some third factor(s). For example, the length of the right thumb and the left thumb show probably a near-perfect rank correlation. Why? Not because your left thumb causes your right thumb's growth. Instead, likely because they are caused by the same factors: genetics + nutrition + physical exercise.
BTW, Pearson correlations don't prove causality either. It's experimentation (&/or a persuasive theoretical model) that is necessary for explaining why the correlation might be so high.
Simply that there is a perfect association, so that a person's rank on one variable can be used to predict precisely what that person's rank on another would be. While correlations can never be used to establish causality, a correlation that strong suggests either the two variables are causally related or very tightly related to some third factor(s). For example, the length of the right thumb and the left thumb show probably a near-perfect rank correlation. Why? Not because your left thumb causes your right thumb's growth. Instead, likely because they are caused by the same factors: genetics + nutrition + physical exercise.
BTW, Pearson correlations don't prove causality either. It's experimentation (&/or a persuasive theoretical model) that is necessary for explaining why the correlation might be so high.
Hi,
First, I think you might be confused about "number of samples" and "correlation". The number of samples being the same is not the same as having correlation. The rho value WOULD be the correlation coefficient, which would tell you how strong a correlation is. I can't tell from your writing whether or not you realize these are different values, but first, know that they are.
What the correlation coefficient (rho in this case) will tell you is how related the two variables are. If you have a perfect correlation (rho of 1; which is highly unlikely for data taken from the natural world) that means that as variable A goes up, variable B will go up by the exact same amount. For example, if A = height, measured in centimeters, and B = weight, measured in kilograms, than for ALL members of your sample population, every change in height by +1 cm will ALWAYS mean +1 change in kilograms. (to use an example from biology)
The spearman's and perason's both give you correlation, so in essence, they are interpreted the same way (so your thinking about pearson's is right, but it seems you would be wrong in thinking that spearman's does something different). The difference between spearman's and pearson's is that pearson's is used for ordinal data rather than continuous data. Ordinal data is either non-parametric or not measured on a completely objective scale (for example, Likert-scale style psychometric questions, subjective feelings about how much pain a patient is experiencing, etc.). The basic premise is the same for these two tests (a positive correlation means that as one variable goes up, the other tends to go up as well). However, pearson's works a bit differently mathematically to control for the fact that at least one of the variables isn't measured on a continuous scale.
Does this make sense?
Ryan Spring I think that you mean that the Spearman is for ordinal (rank), and the Pearson for interval. Right?
Rhianon Allen Ooops! Yes, I wrote that backwards! Camilla - please take note of this important point brought up by Rhianon Allen
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation