Analyzing a Subset of Census Tracts

Most researchers can extract data from the 1980 or 1990 STF1 or STF3 files for the entire state or for an individual county, but if one has a list of particular census tracts, it is not always easy to aggregate the data for just those tracts.

These macros allow one to choose STF1 or STF3 (1990 or 1980 Census) and to name the county and particular tracts of interest. The chosen macro then extracts the data from the RCF-based raw data files, aggregates the tract data (and stores it in a single observation), and saves the state, county, tract and aggregated-tract data into a SAS data set. By default, all variables from the STF1 or STF3 files are saved, but the program also allows the patron to specify a subset of variables.

1990 Census

• Tract Subset Macro -- Save As tractsub90.mac
  
• Tract Subset Program -- Save As tractsub90.sas

1980 Census

• Tract Subset Macro -- Save As tractsub80.mac
  
• Tract Subset Program -- Save As tractsub80.sas

NOTE: The programs and macros discussed here are provided without warranty or guarantees of accuracy or completeness, though every attempt has been made to attain these qualities. It is the researcher's responsibility to verify the results to her or his satisfaction.

No changes are necessary for the Macros (tractsubYY.mac). The programs (tractsubYY.sas) use their respective macros As-Is. Note: When downloading, you should store both the macro and the program in the same directory where you will be invoking SAS.

In the program (tractsubYY.sas) you will need to specify only a few choices (using an editor like emacs or pico):

• Census Data Base you wish to use (STF1 or STF3)
  
  The line in the downloaded program (tractsubYY.sas) looks like:

  %let ver=%str(stf1);

  Your Change: for STF3, you'd change the line to look like:

  %let ver=%str(stf3);

County where your tracts are located

The line in the downloaded program (tractsubYY.sas) looks like:

%let county=%str('037');

Your Change: For Kern County (CNTY value '029'), the line would look like:

%let county=%str('029');

Specific tracts for your chosen subset

The line in the downloaded program (tractsubYY.sas) looks like:

%let traclist=%str('2352.02','2364','2377','2407');

Your Change: If you preferred to include tracts 2381, 2395, and 2405, you'd change the line to look like:

%let traclist=%str('2381','2395','2405');

A few less commonly-needed choices also can be made, if you wish.
• You can specify the destination for your SAS transport and (optionally) data files (By default [i.e., when no changes are made to this line in the program], everything will be saved in /tmp)

  Example: to have everything deposited in your home directory (~/) the line would look like this:

  %let outpath=%str(~/);

• You can exclude or selectively include certain observations using IF statements as explained in the programs themsevles.

• You can specify a subset of your choice of variables, if you wish. Examples are given in the programs.

• You can have the program create a SAS data set (in addition to the SAS Transport File that is created by default) if you wish. The programs instruct you which lines to 'uncomment' (and how to 'uncomment' them), if you want a SAS data set.

• The programs use PROC PRINT to print out all the data that have been saved, including the aggregated line. The data for the entire State will have the usual SUMLEV value of 40 (SUMRYLVL 4 in the 1980 Census), the data for the County will have its usual SUMLEV value of 50 (SUMRYLVL 11 in the 1980 Census), and each separate Census Tract will have a SUMLEV value of 80 (SUMRYLVL 14 in the 1980 Census). The aggregated data will be in a single observation, with the 'assigned' SUMLEV value of 80.9 (SUMRYLVL 14.9 in the 1980 Census).

Researchers' Notes:
Click here for detailed discussion of how the Tract Subset macros work
• A number of variables represent median values or per capita values. The chosen macro (tractsubYY.mac) automatically averages these kinds of variables over the chosen Census Tracts, so that the value deposited in the Aggregated line (the observation in the SAS data set with the SUMLEV value of 80.9 for the 1990 Census or a SUMRYLVL value of 14.9 for the 1980) is an average, not a sum. Click here for a detailed explanation and lists of variable names.

• Some variables have text or other information specific to a Census Tract that cannot be averaged or summed. The macro sets these kinds of variables to missing or labels them as 'Custom'.

  Examples:

  Variable  Purpose            Text entered in 80.9 (or 14.9) by macro:
  ---------------------------------------------------------------------
  TRACTBNA  Tract Number       "Custom"
  ANPSADPI  Descriptive Text   "Custom Tract Grouping"
  LONGITUDE Longitude          (nothing)

  Click here for details.