Background & History | Documentation
Geographic coverage | Directory structure/file naming conventions
People

Background and history:

The 1970 Fifth Count Special Tabulation was purchased in 1975 by the U.S. Army Corps of Engineers for their civil works planning process and placed in a national information system at Lawrence Berkeley National Laboratory (LBL). The original data were stored on 9-track tapes for retrieval on the LBL Control Data supercomputers. For space saving purposes a special binary compression algorithm was written tuned to the characteristics of that computer (60 bit word, 6 bit CDC character set). In 1989 the data were converted to a computer-independent binary compression format using run-length encoding. The data were then installed in the LBL Socio-Economic-Environmental Demographic Information System (SEEDIS) running on a DEC VAX computer running the VMS operating system. The last known SEEDIS computer is currently running internally at the U.S. Census Bureau.

During the year 2000-2001 a rescue effort for this data file was mounted by the University of California Library and the U.C. Data Archive and Technical Assistance (UC DATA). This involved receiving data on DLT tape from the Census Bureau and writing decompression software which would run on Unix machines with the Sun Microsystems SOLARIS operating system. This process was completed (except for a few details noted below) in May 2001.

Documentation:

The file 1970-Fifth-Count.doc consists of a machine-readable version of the original Census Bureau documentation for the original Zipcode data file. It can be used for purposes of understanding sampling and data suppression as well as the multidimensional nature of the summary tabulations. IT CANNOT BE USED AS A DATA DICTIONARY TO READ THESE FILES, BECAUSE IT WAS ORIENTED TOWARD TAPE FILES IN IBM FORMAT.

The file cens5th.hel.txt consists of portions of the original LBL SEEDIS documentation of the data. It has labels for each particular variable (cell) in the record. The file
      cens5th-spss-setup.sps
      cens5th-s06-mcd-spss-test.txt
are, respectively, an SPSS setup file which can read the file and do basic statistics on all variables and the output from an SPSS run using the setup files for MCD records of California (State FIPS code 06). The SPSS file identifies the field positions for each variable in a data file.

Geographic coverage:

The data consists of Fifth Count summary tabulations consistent with the 1970 Fifth Count Zipcode area tabulations of the regular census file. The data are available for Minor Civil Divisions (level MCD70) which are usually judicial subdivisions of counties, Census Tracts (level TRACT) corresponding to 1970 Census Tract boundary maps, and Census Block Groups and Enumeration Districts (level BGED70).

Block Groups are subunits of Census Tracts in urban areas, while Enumeration Districts are subunits of counties for rural counties or rural portions of urban counties. BGED70 records generally correspond to areas which contained about 250 housing units. Level TRACT was only provided for urban counties which were assigned Census Tracts in 1970 (not until the 1990 census was the entire country assigned census tracts by the Bureau). This level has covers only 465 counties. Level MCD70 was only tabulated for rural counties, or 2568 counties. Thus the two levels are complementary in their coverage of the entire country. Level BGED70 covers 3141 counties of the 50 states and the District of Columbia. One county was lost in the original processing and three counties were not decompressed because of processing problems. These counties, from the the state of Michigan (s26), are:
            c091 Lenawee
            c117 Montcalm
            c121 Muskegon

The file cens5th-county.cod is a self-describing data file from the SEEDIS system which tells for each county in the United States (according to Census 1970 geography) whether and how many records (areas) of each level of geography (MCD70, TRACT, BGED70) are found in that county. The data portion has the following format:

    Column           Content
      1        FIPS state and county code 
      2        Number of BGED70 records
      3        flag =1 if BGED70 records exist
      4        Number of TRACT records
      5        flag =1 if TRACT records exist
      6        Number of MCD70 records
      7        flag =1 if MCD70 records exist

Directory structure and file naming conventions:

The directory and data file structure is <level of geography>/<state code>/<county code>.dat
where <level of geography> is MCD70 or TRACT or BGED70 and <state code> is of the form snn where nn is the FIPS (Federal Information Processing Standard) 2 digit code for that state. and <county code> is in the form cnnn where nnn is the FIPS code for that county (unique within the state). Thus
            mcd70/s06/c023.dat
contains the Minor Civil Division (MCD70) level of geography data for Humboldt County (c023) California (s06). For a list of FIPS codes, go to the web site
http://oseda.missouri.edu/plue/geocorr/htmls/counties.html

People:

This project was a cooperative project between the University Library and UC DATA. Original data files were reclaimed from the LBL SEEDIS system with the help of Deane W. Merrill of LBL (retired) and Chris Stuber of the Census Bureau. Unpacking and decompression were done by Sungman Cho, Fredric Gey, and Jon Stiles of UC DATA using software programmed by Harvard H. Holmes of LBL. Creation of the electronic version of Census Bureau documentation was done by Jon Stiles and Ilona Einowski of UC DATA. Fred Gey adapted the LBL documentation file, reducing it's size by separating out the geography files. Jon Stiles created the SPSS setup file, and Fred Gey ran the test output.

Document maintained on sunsite.berkeley.edu
FCG 6/13/01