Home Statistics and Visualization R-bloggers Can we trust the correlation coefficient in designed field experiments?