Transforming Multiple-record Data into Single-record Format When Number of Variables Is Large

In one large survey project each record in the raw data set contains information from one survey form. At some point, Adenormalization@ of the data needs to be done transposing a multiple-record-per-object data set to a one-record-perobject data set that retains all raw variables from all data sources. The large dimension of the original data set requires an approach that can: a) carry out the denormalization in a reasonable amount of time; b) handle the indexing and standardization of variables= names automatically; c) easily adapt as the set of variables changes over time. One approach uses a two-step PROC TRANSPOSE. This approach deals elegantly with issues b) and c) , but it takes a substantial amount of time to run. The second approach uses a driving macro based on the RETAIN statement. Although this method requires some maintenance, it runs much more quickly than the TRANSPOSE method.