if i want to test association of two continuous variables but the sample size is four, can i run pearson correlation or i have to go non-parametric with spearman ?
The choice between Pearson and Spearman is not a question of sample size, but whether you want to quantify and test linear or monotonic association.This might relate to an underlying assumptions on your variables. I would echo Mahadi Hasan Miraz in that as the sample size is so small, I am unsure if numerical tests of association would produce much value for your research. A simple scatter plot of the points might be even more convincing for a reader in this case.
If your both the variables are continuous then you can go for pearson's correlation and sample size is 4 which give the value but if possible then use more no. Of observation to improve results.