STAT 20 Lecture 28: stats 20 10-24-18

• School: Berkeley
• Department: Statistics
• Course: STAT 20
• Professor: Fletcher Ibser

Document Summary

Different kinds of scatter plots: others are points that are fat away from most of the data. Outliers need to be considered as long as they are feasible: nonlinear scatter plot has some pattern that doesn"t follow a line. What we know from a correlation is whether a linear relationship exists. There are non-linear relationships and non-linear scatter plots: football shaped, linear, no outliers, x and y follow normal curve, homoscedastic (same spread in vertical strips, ecological correlation tends to overstate relationships for individuals. Correlation does not mean causation, but one may start growing conclusions with many related correlations. For example, pipe smokers inhale smoke very deeply and tend to get cancer lower in the lungs. Cigar smokers only do light puffs and do not inhale deeply and they cancer in their upper respiratory system. Cigarette smokers tend to get cancer all over. People who do chewing tobacco get cancer in their mouth.