8.7 DO Loops

Sometimes blocks of script items need to be repeated several times. An example of this is asking the same set of questions about each member of a household. Although it is quite possible to do this by using multiple copies of the blocks of items, DO loops allow the script to be drastically simplified.

A DO loop is begun by an item of type DO. This is followed by a series of other item types. The loop is ended by a ENDD item. One of the parameters to the DO item specifies the maximum number of times the DO loop executes. Once this is exceeded the ENDD item becomes unprotected and is then the current item.

Some special conditions apply to DO loops:

• A DO loop cannot be nested inside another DO loop.
• Any item type is allowed inside the DO loop except TITL, LINK, and SCAL.
• The names of the DO and ENDD items as well as all items within the loop must be no more than 6 characters in length.
• The SPS of the first item must refer to the last item in its SPS. In other words, the item immediately following the DO item must refer to the item immediately preceding the ENDD item.

The compiler deals with loops by expanding them into a repeated series of items. Suppose the DO loop shown in Table 8 is to be repeated 3 times.

Table 8: List of items within a DO loop with associated SPSs.

Type	Variable Name	SPS
DO	LOOPA
INFO	Q1	Q4[-1]$>$.
CHCE	Q2	Q1[0]=1
CHCE	Q3	Q2[0]$>$.
CHCE	Q4	Q3[0]$>$.
ENDD	ENDLOOPA	Q44$>$.

The compiler will expand the loop as shown in Table 9.

When expanding the loop items within it are duplicated and renamed. The compiler also alters the SPSs for the repeated items (see Table 9). Once expanded, items within the loop are treated as any other item by the compiler.

Table 9: List of items within an expanded DO loop. The associated SPSs have been automatically modified by GEIS. Secondary variables LOOPA2 and LOOPA3 have been created.

Type	Variable Name	SPS
DO	LOOPA
	LOOPA2
	LOOPA3
INFO	Q11	LOOPA=1
CHCE	Q21	Q11=1
CHCE	Q31	Q21$>$.
CHCE	Q41	Q31$>$.
INFO	Q12	Q41$>$.
CHCE	Q22	Q12=1
CHCE	Q32	Q22$>$.
CHCE	Q42	Q32$>$.
INFO	Q13	Q42$>$.
CHCE	Q23	Q13=1
CHCE	Q33	Q23$>$.
CHCE	Q43	Q33$>$.
ENDD	ENDLOOPA	Q43$>$.

The DO item itself generates 3 variables to control the looping. The DO item generates 3 variables within the ANSWERS data set: loopA, loopA2, and loopA3. These are used to control the looping.

If the item is called QNAME, the variables will be named QNAME1, QNAME2, and QNAME3. The variable QNAME1 will contain the highest loop count, the variable QNAME2 will contain the maximum possible loop count, and the variable QNAME3 will contain the current loop count. These variables can be directly referenced within the script.

The script syntax for the SPSs within the loop is the same as normal except that square brackets enclosing an integer are used to indicate the loop count. The square brackets follow an item name for an item that appears within a loop. They are used to reference the values of items within iterations around the loop.

A positive integer within square brackets (e.g. Q2[3]) refers to the item's value on that particular iteration. A negative integer (e.g. Q2[-5]) is used to refer to the item's value on an iteration relative to the current one. A zero integers (e.g. Q2[0]) refers to the item's value on the current iteration. This is also shown in Table 8. For example, a SPS Q2[-1]=3, which would be evaluated to be true in an interview when the answer to Q2 on the previous iteration around the loop was `3'.

Table 10: Interpretation of positive, zero, and negative indices in self-protection statements within a DO loop.

Index	Meaning
k>0	Item value on the $k$th loop
k=0	Item value on current loop
k<0	Item value $k$ iterations earlier

Some special conditions apply to square bracket indices:

• The ENDD item SPS can only have positive integers within square brackets.
• Items within the loop can only have 0 or negative numbers.
• The first item in a loop normally uses negative integers to reference the last item in the loop on the previous time round the loop.

These points are also shown in Table 10.

As explained above, positive indices correspond refer absolutely to a particular iteration. Thus, Q1[3] means the value of Q1 on the 3rd iteration. Negative indices refer in a relative sense to earlier iterations. Thus, Q1[-3] means the value of Q1 3 iterations ago.

It is possible to use indices that do not correspond to a loop iteration. If the maximum number of iterations is 5 then the SPS Q1[7] or Q1[-7] would be incorrect. The compiler will report the SPS Q1[7] as a syntax error. However, the Q1[-7] would be accepted, but the SPS evaluates it as a missing value during the interview.

If a negative index is used, such as Q1[-3], then the item it refers to will not exist until the fourth iteration. In this case the compiler will replace the reference by a missing value. The evaluation of the SPS Q1[-3]=1 and DO3$>=$3 with successive loop iterations in shown in Table 11.

Table 11: Example of the evaluation of an SPS that uses a relative index through successive iterations within a loop. The SPS in this case is: Q1[-3]=1 and DO3$>=$3.

Iteration	SPS
1	.=1 and DO3$>=$3
2	.=1 and DO3$>=$3
3	Q11=1 and DO3$>=$3
4	Q12=1 and DO3$>=$3
5	Q13=1 and DO3$>=$3

An internal macro variable called &CURRITER is available that holds the current loop count. So ``Q&CURRITER'' becomes ``Q1'' on the first time round the loop, then ``Q2'' the second time. This can be used to reference loop items as well as external items. The macro variable is illustrated in Table 12.

Table 12: Example of the use of the CURRITER macro variable within a DO loop.

Type	Name	SPS
CHCE	Num
DO	LOOPA	Num$>$.
info	Q1	Q4[-1]$>$. and LOOPA3$<=$Num
chce	Q2	Q1[0]=1 and LOOPA3$<=$Num
chce	Q3	Q2[0]$>$. and LOOPA3$<=$Num
chce	Q4	Q3[0]$>$. and LOOPA3$<=$Num
ENDD	endloopa	Q2&CURRITER$>$. and LOOPA3$>=$NUM

A loop is always set to execute a maximum number of times. If a loop is supposed to be exited before the maximum loop count is reached then the SPS for the ENDD must be set accordingly. The example in Table 8 could have used: (Q41=1) or (Q42=1) or (Q43=1) or (Q44=1), which would allow the loop to be exited from any of the four iterations as long as the answer to Q4 was code 1.

Changing the texts for questions on successive iterations can be done by using answer-quoting¹⁵. One way this could be done is to quote the loop count as in the question:

``How old is person ^LOOPA3^ in your household?''

This appears in consecutive loops as:

``How old is person 1 in your household?''

``How old is person 2 in your household?''

``How old is person 3 in your household?''

 

Alternatively, a SCAL item can be used to supply the question texts. Suppose the SCAL contains the text to be quoted as four character strings. These strings can be quoted by using the loop count to refer to a particular SCAL item value. Suppose the SCAL item contained the 4 texts: ``man'', ``woman'', ``boy'', and ``girl'' and that a loop that refers to it runs 4 times. If the single question in the script is:

``Is person ^LOOPA3^ in your household a ^SCAL[0]^?''

then the following questions will appear as the loop iterates:

``Is person 1 in your household a man?''

``Is person 1 in your household a woman?''

``Is person 1 in your household a boy?''

``Is person 1 in your household a girl?''

``Is person 2 in your household a man?''

``Is person 2 in your household a woman?''

``Is person 2 in your household a boy?''

``Is person 2 in your household a girl?''

``Is person 3 in your household a man?''

``Is person 3 in your household a woman?''

``Is person 3 in your household a boy?''

``Is person 3 in your household a girl?''

``Is person 4 in your household a man?''

``Is person 4 in your household a woman?''

``Is person 4 in your household a boy?''

``Is person 4 in your household a girl?''