When executing a dataset, how many records are typically read?

When working with datasets in SAS, typically, one record is processed at a time during the execution of a DATA step or similar operations. This approach aligns with the iterative nature of how SAS reads data. During the execution, SAS reads each record sequentially from the dataset, allowing for detailed manipulation and processing on a record-by-record basis.

This model of reading one record enhances the flexibility of data manipulation, as it allows for computations and data transformations to be applied to individual records before moving on to the next. This mechanism facilitates tasks such as conditional processing, creating new variables, or filtering records based on specific criteria.

In the context of how many records are read, saying that only one record is read at a time reflects the fundamental functioning of SAS during data step processing. Other options don't accurately represent the typical operation. For instance, while it's theoretically possible for zero records to be read if a dataset is empty or does not exist, or for multiple records to be processed collectively in a procedure step, these scenarios are not the norm for the typical execution of a DATA step. Additionally, the size of the dataset does influence performance and efficiency but does not change the fundamental sequential reading of one record at a time.