##
Overview
Section* *

In this lesson, we investigate statistical analyses that are typically performed when dealing with two or more continuous numeric variables. Specifically, we investigate:

- the
**GPLOT**procedure, to create publication quality*x-y*scatter plots of any two numeric variables in a SAS data set - the
**CORR**procedure, to compute various correlation coefficients between two or more numeric variables in a SAS data set - the
**REG**procedure, to perform a regression analysis on any subset of numeric variables in a SAS data set

## Objectives

Upon completion of this lesson, you should be able to:

Upon completing this lesson, you should be able to do the following:

- use the CORR procedure to tell SAS to calculate Pearson correlation coefficients among a set of numeric variables
- use the CORR procedure's SPEARMAN, KENDALL, and HOEFFDING options to tell SAS to calculate alternative coefficients
- read typical correlation procedure output in order to be able to extract the calculated correlations and their associated
*P*-values - use the CORR procedure's WITH statement to tell SAS to calculate only the correlation coefficients among the variables in the WITH and VAR statements
- understand how sample size can affect the significance of a correlation coefficient
- interpret a correlation coefficient
- use the CORR procedure's PARTIAL statement to tell SAS to calculate partial correlations among variables
- use the CORR procedure's BEST =
*n*option to tell SAS to print only the first*n*of the ordered estimated correlations - use the REG procedure to compute a regression equation between two numeric variables
- use the REG procedure's MODEL statement to tell SAS which variable to treat as the response variable and which variable to treat as the predictor variables
- read the typical SAS output from regression analysis to pull off key information, such as parameter estimates, confidence intervals, and
*P*-values - use the REG procedure's PLOT statement to request residual diagnostic plots
- use the GPLOT procedure to request plots containing estimated regression equations, 95% confidence intervals about the mean of
*y*, and 95% prediction intervals about the individual*y*-values - use the REG procedure to conduct a regression analysis involving quadratic terms
- use the REG procedure to conduct a regression analysis involving transformed variables

##
Textbook Reference
Section* *

Chapter 5 of the textbook.