What does the SET statement do in a DATA step?

The SET statement in a DATA step is primarily used to read observations from existing datasets into a new dataset. When the SET statement is invoked, it allows the user to access the variables of one or more existing datasets and create a new dataset that contains their data. This can involve processing and manipulating the data as specified within the DATA step while retrieving the data from the source dataset(s).

For example, if a programmer needs to combine or analyze data stored in various datasets, the SET statement serves as a crucial tool for pulling these observations directly into the DATA step environment. After this, additional programming logic (such as IF statements, calculations, or transformations) can be applied to those observations as they are being read.

The other options do not accurately describe the primary function of the SET statement. While creating a new dataset or merging datasets could be implicit outcomes of using SET, the statement itself does not create a dataset nor directly modify formats. The merging of datasets would typically be accomplished through other statements or methods in SAS, rather than through the SET statement alone.