Program States

To run a program session, you need to define your active dataset, transform the data, and then analyze it. This order conforms very closely to the order the program must follow as it processes your commands. Specifically, the program checks command order according to the program state through which it passes. The program state is a characteristic of the program before and after a command is encountered. There are four program states. Each session starts in the initial state, followed by the input program state, the transformation state, and the procedure state. The four program states in turn enable the program to set up the environment, read data, modify data, and execute a procedure. The figure shows how the program moves through these states. The program determines the current state from the commands that it has already encountered and then identifies which commands are allowed in that state.

A session must go through initial, input program, and procedure states to be a complete session. Since all sessions start in the initial state, you need to be concerned primarily with what commands you need to define your active dataset and to analyze the data. The following commands define a very minimal session:

GET FILE=DATAIN.
FREQUENCIES VARIABLES=ALL.

The GET command defines the active dataset and the FREQUENCIES command reads the data file and analyzes it. Thus, the program goes through the required three states: initial, input, and procedure.

Typically, a session also goes through the transformation state, but it can be skipped as shown in the example above and in the diagram in the preceding figure. Consider the following example:

TITLE 'PLOT FOR COLLEGE SURVEY'.

DATA LIST FILE=TESTDATA
 /AGE 1-3 ITEM1 TO ITEM3 5-10.

VARIABLE LABELS ITEM1 'Opinion on level of defense spending'
  ITEM2 'Opinion on level of welfare spending'
  ITEM3 'Opinion on level of health spending'.
VALUE LABELS ITEM1 TO ITEM3 -1 'Disagree' 0 'No opinion' 1 'Agree'.
MISSING VALUES AGE(-99,-98) ITEM1 TO ITEM3 (9).
RECODE ITEM1 TO ITEM3 (0=1) (1=0) (2=-1) (9=9) (ELSE=SYSMIS).
RECODE AGE (MISSING=9) (18 THRU HI=1) (LO THRU 18=0) INTO VOTER.
PRINT /$CASENUM 1-2 AGE 4-6 VOTER 8-10.
VALUE LABELS VOTER 0 'Under 18' 1 '18 or over'.
MISSING VALUES VOTER (9).
PRINT FORMATS VOTER (F1.0).

FREQUENCIES VARIABLES=VOTER, ITEM1 TO ITEM3.

The program starts in the initial state, where it processes the TITLE command. It then moves into the input state upon encountering the DATA LIST command. The program can then move into either the transformation or procedure state once the DATA LIST command has been processed.

In this example, the program remains in the transformation state after processing each of the commands from VARIABLE LABELS through PRINT FORMATS. The program then moves into the procedure state to process the FREQUENCIES command. As shown in the preceding figure, the program can repeat the procedure state if it encounters a second procedure. The program can return to the transformation state if it encounters additional transformation commands following the first procedure. Finally, in some sessions the program can return to the input program state when it encounters commands such as FILE TYPE or MATCH FILES.