Defined:
proc anova [data=dataset] [outstat=outputdataset]; model dep ... [= indep ...] [ / [int] [nouni]]; class classvars ... ; [means vars ... [ / [bon] [duncan] [scheffe] [t] [tukey] [waller] [clm] [alpha=prob] [e=effect]]; ] [run;] [quit;]
Notes
[data=dataset] - dataset is the name of the data set you want analysed. If no data set is named, SAS uses the most recently created set.
[outstat=outputdataset] - outputdataset is a data set name for listing statistics such as sum of squares, F values, and probability levels.
model dep ... - A model MUST be defined for anova to work. dep is the list of variables which are the dependant in the model.
[= indep ...] - Lists the independant variables. If none is entered, SAS performs the test that the mean of dep... is 0.
[int] - Int forces proc anova to consider the intercept as an effect in the model.
[nouni] - Suppresses the univariate statictics usually outputted by proc anova.
class classvars ... - A class statement must be given. classvars is a list of variables as classification variables, or subgroups.
[means vars ... ] - Computes the means and tests them for each of the variable listed.
[bon] - Perforns the Bonferroni t-tests
[duncan] - Performs Duncan's multiple range test
[scheffe] - Performs Scheffe's multiple comparison test
[t] - Performs a pairwise t-test on the variables
[tukey] - Performs Tukey's test
[waller] - Performs Waller-Duncan multiple tests
[clm] - Gives output from the BON, SCHEFFE, and T in one
[alpha=prob] - Sets the alpha level for your tests. Prob can be between .0001 and .9999. By default, alpha is .05. Also, if you perform Duncan's tests, alpha must be either .01, .05, or .10.
[e=effect] - Defines which error means squares to use in testing. By default, SA uses the residual mean squares error.
[run;] - Tells SAS to compile to memory
[quit;] - Tells SAS to quit using proc anova