Since almost all of the points lie flat near ây=0â for âxâ¤1, then only one or two points drift upwardâand one extreme outlierâthere isnât a single straight line that captures the bulk of the data. Most points would lie far from the regression line youâd draw through that big spike.
In other words, the scatter doesnât hug a line very tightly, so the correlation coefficient ârâ will be small in magnitude.