The SURVEYMEANS Procedure

POSTSTRATA Statement

  • POSTSTRATA variables / PSTOTAL= <option>;

  • POSTSTRATA variables / PSPCT= <option>;

The POSTSTRATA statement names variables that form the poststrata to adjust the sampling weight for analyzing the survey. The combinations of categories of POSTSTRATA variables define the poststrata in the sample.

The POSTSTRATA variables are one or more variables in the DATA= input data set. These variables can be either character or numeric. The formatted values of the POSTSTRATA variables determine the categorical levels. Thus, you can use formats to group values into levels. See the FORMAT procedure in the Base SAS Procedures Guide and the FORMAT statement and SAS formats in SAS Formats and Informats: Reference for more information.

You must specify either poststratification totals or poststratification proportions, but not both, in the POSTSTRATA statement after a slash (/).

PSTOTAL=SAS-data-set | values
POSTTOTAL=SAS-data-set | values
PSCONTROL=SAS-data-set | values

specifies an input data set that contains the poststratum totals or specifies the population poststratum totals as positive values. The SURVEYMEANS procedure uses this information to compute weight adjustment for poststratification.

You can provide poststratification totals by specifying PSTOTAL=SAS-data-set, which names a SAS data set that contains the poststratification variables and the poststratum totals. This data set is called the poststratum total data set.

A poststratum total data set must contain all the poststratification variables that are listed in the POSTSTRATA statement and all the variables listed in the BY statement. If there are formats associated with the POSTSTRATA variables and the BY variables, then the formats in the poststratum total data set for these variables must be consistent with those in the DATA= data set in the PROC SURVEYMEANS statement.

A poststratum total data set must have a variable named _PSTOTAL_ that contains the poststratum totals. The value of _PSTOTAL_ must be positive.

When poststratum levels are easy to identify, their corresponding poststratum totals can be specified as a list of positive numbers.

ORDER=FORMATTED is used to order the levels of poststratum levels.

You can separate values in the PSTOTAL= option with blanks or commas. The number of values must equal the number of poststrata in the data. List the values in the order of the corresponding poststratum level. The sum of the values should equal the total sampling weights.

PSPCT=SAS-data-set | values
POSTPCT=SAS-data-set | values

specifies an input data set that contains the population poststratum proportions or specifies the population poststratum proportions as positive values. The SURVEYMEANS procedure uses this information to compute weight adjustment for poststratification.

You can provide poststratification proportions by specifying POSTPCT=SAS-data-set, which names a SAS data set that contains the poststratification variables and the poststratum poststratification. This data set is called the poststratum proportion data set.

A poststratum proportion data set must contain all the poststratification variables that are listed in the POSTSTRATA statement and all the variables listed in the BY statement. If there are formats associated with the POSTSTRATA variables and the BY variables, then the formats in the poststratum proportion data set for these variables must be consistent with those in the DATA= data set in the PROC SURVEYMEANS statement.

A poststratum proportion data set must have a variable named _PSPCT_ that contains the poststratum proportions. The value of _PSPCT_ must be positive.

You can provide poststratum proportions either as positive decimal numbers between 0 and 1 for all poststrata or as positive percentages that must be less than 100 for all poststrata. If any value of the proportions is greater than 1, the procedure treats all proportions as percentages instead of decimal numbers.

When poststratum levels are easy to identify, their corresponding poststratum proportions can be specified as a list of positive numbers.

ORDER=FORMATTED is used to order the levels of poststratum levels.

You can separate values in the POSTPCT= option with blanks or commas. The number of values must equal the number of poststrata in the data. List the values in the order of the corresponding poststratum level.

If you provide the proportions as decimal numbers, then the sum of these values over all poststrata must be 1.

If you provide the proportions as percentages, then the sum of these percentages over all poststrata must be 100.

You can also optionally specify the following option to create an output data set to store the poststratification weights.

OUTPSWGT=SAS-data-set
OUT=SAS-data-set

names a SAS data set to contain poststratification weights. For information about poststratification weights, see the section Poststratification.

If you also specify an OUTWEIGHTS= method-option for VARMETHOD=BRR or VARMETHOD=JACKKNIFE in the PROC SURVEYMEANS statement , the OUTPSWGT= option is ignored. The poststratification weights for the full sample and the replication weights adjusted for poststratification are stored in the OUTWEIGHTS= data set.

For more information about the contents of the OUTPSWGT= data set, see the section Poststratification Weights Output Data Set. For more information about the contents of the OUTWEIGHTS= data set, see the section Replicate Weights Output Data Set.