Monday, Oct 13
Data Center | Packaged Main Data/Doc | Packaged Supplemental Data/Doc | CDS Doc | FIMS | Data News | Summary Lists

                    A Panel Study of Income Dynamics:
               1968-1980 Retrospective Occupation-Industry
                           Files Documentation

                               (Release 1)












   Conducted with grants from the National Institute on Aging and the
                        National Science Foundation.













                         Survey Research Center

                      Institute for Social Research
                       The University of Michigan
                           Ann Arbor, Michigan

                                  1999








                            TABLE OF CONTENTS



Section I: Overview
       1968-1980 Retrospective Occupation-Industry Files
       Persons For Whom the Data are Available
       Background for the Retrospective Occupation-Industry Files
       How to Obtain the Files and Whom to Contact with Questions

Section II: Structure of the Files
       Number of Records and Sort Order
       Variables on the 1968-1980 Retrospective Occupation-Industry
       Files

Section III: Reliability Rates and Cautionary Notes

Section IV: Linking Records
       Using the 1968-1980 Retrospective Occupation-Industry Files
       with Cross-Year Individual Files
       Using the 1968-1980 Retrospective Occupation-Industry Files
       with the Family Files



Section I: Overview


1968-1980 Retrospective Occupation-Industry Files

The set of 1968-1980 Retrospective Occupation-Industry Files is
comprised of thirteen family-level files, one for each wave,
containing three-digit 1970 Census occupation and industry codes
for selected PSID Heads and Wives.

The PSID used a one-digit occupation code, and later a two-digit,
until 1981 when the three-digit 1970 Census code became standard.
The complete history of PSID occupation and industry coding is
described in fuller detail below.

Records on these files are short and contain only the relevant
identifier for each year's family.  In addition to the identifiers,
each file has two sets of variables, one pair of three-digit
occupation and industry variables for the current Head and a second
for the current Wife.

The 1968-1980 Retrospective Occupation-Industry Files are designed
to be linked to the relevant 1968 through 1980 Family Files, and
family identifiers are compatible with those datasets.


Persons For Whom the Data are Available

The occupations and industries were recoded from original reports
for Heads and Wives who met the following criteria:

       a)     Original sample Heads and Wives/"Wives still living by
              1992 who reported main jobs in at least three waves
              during the period 1968-1992, with at least one of those
              reports prior to 1980.

       b)     Additionally, original sample Heads and Wives/"Wives" who
              had reported at least one main job between 1968 and 1980
              but were known to have died by 1992.

The selection criteria did not include all Heads and Wives/"Wives"
who had worked between 1968 and 1980.  Those who were still living
but had reported only one or two jobs during the period of interest
were excluded, as were all nonsample Heads and Wives/"Wives".
Therefore not every Head or Wife/"Wife" has data on these files.


Background for the Retrospective Occupation-Industry Files

The 1968-1980 Retrospective Occupation-Industry Files originated as
part of the Working Lives and Mortality in an Aging National Cohort
project, funded by the National Institute on Aging and with
Benjamin Amick of The Health Institute at the New England Medical
Center and David R. Williams of the Survey Research Center here at
ISR.  This project required three-digit occupation and industry
codes throughout the course of the PSID.

Between 1968 and 1980, the PSID had coded occupations and
industries using various combinations of one- and two-digit codes.
Occupations were coded using the one-digit code from 1968 through
1975.  Classic industry questions ("What kind of business or
industry is that?") were not asked until 1971, when a two-digit
industry code was utilized.  Two-digit occupations were coded from
1976 through 1980, and in one wave, 1974, the three-digit Census
code was used along with the one-digit.

The table below shows what was coded in which years for whom,
depending on employment status and whether the person was a Head or
a Wife.

                                  NUMBER OF     NUMBER OF
YEAR AND            QUESTION      DIGITS FOR    DIGITS FOR
EMPLOYMENT STATUS   REFERENCE     OCCUPATION    INDUSTRY

1968                                  1         Not coded
 Head Employed        F2-F3
 Head Unemployed      G1*
 Head Out of Labor
   Force              H2*
 Wife                 I10*

1969                                  1         Not coded
 Head Employed        D1-D2
 Head Unemployed      **
 Head Out of Labor
   Force              F2*
 Wife                 G3*

1970                                  1         Not coded
 Head Employed        D2-D3
 Head Unemployed      E6*
 Head Out of Labor
   Force              F3*
 Wife                 G3*

1971                                  1            2
 Head Employed        D2-D3,D3a
 Head Unemployed      E6,E6a*
 Head Out of Labor
   Force              F3,F3a*
 Wife                 G3,G3a*

1972                                  1            2
 Head Employed        D2-D3,D3a
 Head Unemployed      E6,E6a*
 Head Out of Labor
   Force              F3,F3a*
 Wife                 G3,G3a*

1973                                  1            2
 Head Employed        D2-D3,D3a
 Head Unemployed      E6,E6a*
 Head Out of Labor
   Force              F3,F3a*
 Wife                 G3,G4*

1974                                 1 and 3       2
 Head Employed        D2-D3,D4
 Head Unemployed      E6,E7*
 Head Out of Labor
   Force              F3,F4*
 Wife                 G3,G4*

1975                                   1           2
 Head Employed        D2-D3,D4
 Head Unemployed      E6,E7*
 Head Out of Labor
   Force              F3,F4*
 Wife                 G3,G4*

1976                                   2           2
 Head Employed        D2-D3,D4
 Head Unemployed      E12,E13*
 Head Out of Labor
   Force              F4,F5*
 Wife                 G3,G4*

1977                                   2           2
 Head Employed        D2-D3,D4
 Head Unemployed      E15,E16*
 Head Out of Labor
   Force              F6,F7*
 Wife                 G6,G7*

1978                                   2           2
 Head Employed        D2-D3,D4
 Head Unemployed      E27,E28*
 Head Out of Labor
   Force              F16,F17*
 Wife                 G6,G7*

1979                                   2           2
 Head Employed        C6-C7,C8
 Head Unemployed      D13,D14*
 Head Out of Labor
   Force              E4,E5*
 Wife Employed        F7-F8,F9
 Wife Unemployed      G6,G7*
 Wife Out of Labor
   Force              H4,H5*

1980                                   2           2
 Head Employed        C6-C7,C8
 Head Unemployed      D12,D13*
 Head Out of Labor
   Force              E4,E5*
 Wife Employed        F7-F8,F9
 Wife Unemployed      G6,G7*
 Wife Out of Labor
   Force              H4,H5*

 *     This question asks what the Head or Wife did on his or her
       last job or what his or her usual occupation was.  This may or
       may not have been during the prior calendar year.  If the
       prior year's work hours were zero, then this was precoded to
       zero for the retroactive three-digit recoding.

**     Occupation and industry questions were not asked for Heads in
       this category in this wave.


The retrospective coding was begun in October, 1996 using a custom
SAS application.  It necessitated accessing paper materials from
the PSID archives for the written descriptions of the jobs.  To
save time and increase reliability, the coder coded all occupations
and industries for each person across all required years before
moving on to the next case.  A table of the number of occupation-
industry pairs to be coded per person follows:

       NUMBER OF   NUMBER    NUMBER
       PAIRS PER   OF        OF PAIRS
       PERSON      PERSONS   RECODED
          1           628        628
          2           553      1,106
          3           667      2,001
          4           661      2,644
          5           643      3,215
          6           573      3,438
          7           570      3,990
          8           515      4,120
          9           470      4,230
         10           431      4,310
         11           473      5,203
         12           436      5,232
         13         2,166     28,158

TOTAL               8,786     68,275


How to Obtain the Files and Whom to Contact with Questions

The 1968-1980 Retrospective Occupation-Industry Files are available
through the internet.  The URL (Uniform Resource Locator) or
loaction of the PSID home page is:

        /

If you have questions that are not answered by this documentation,
you can contact PSID staff by e-mail:

         psid_staff@isr.umich.edu


Section II: Structure of the File

The Retrospective Occupation-Industry Files are family-level files.
That is, data for Heads and Wives are included on the same records,
just like other PSID family files.

Numbers of Records and Sort Order

The 1968-1980 Retrospective Occupation-Industry Files contain
varying numbers of records, depending on the number of families
with either a Head or a Wife who met the criteria outlined above.
Hence, not every Head or Wife in a given wave was recoded on these
files.  Below is a year-by-year summary of case counts:

            ORIGINAL N   TOTAL    NUMBER       NUMBER
            FROM FAMILY  CASES    OF HEADS     OF WIVES
    YEAR       FILE      CODED    RECODED      RECODED
    1968       4802       4802      3394         1193
    1969       4460       3596      3378         1301
    1970       4645       3870      3644         1408
    1971       4840       3976      3670         1416
    1972       5060       4070      3698         1415
    1973       5285       4177      3737         1383
    1974       5517       4345      3823         1445
    1975       5725       4435      3869         1467
    1976       5862       4485      3839         1526
    1977       6007       4595      3903         1514
    1978       6154       4712      3959         1551
    1979       6373       4939      4059         1673
    1980       6533       5148      4165         1768

Each file is sorted, in ascending order, by the yearly Interview
Number (e.g., for the 1968 file, ID68; for the 1969 file, ID69,
etc.).


Variables on the 1968-1980 Retrospective Occupation-Industry Files

The 1968-1980 Retrospective Occupation-Industry Files contain only
five variables per file.  For detailed information on codes, refer
to the codebooks, but aside from the family identifier, two sets of
occupation and industry codes are included on each file.  The first
set (e.g., on the 1968 file, HDIND68 and HDOCC68) are the three-
digit industry and occupation for the Head, if qualified under the
criteria described in Section I above, in that wave.  The second
set (e.g., on the 1968 file, WFIND68 and WFOCC68) are the three-
digit industry and occupation for the Wife, if there was one and
she qualified for recoding, in that wave.


Section III: Reliability Rates and Cautionary Notes

Occupation and industry coding is a specialized coding task.  The
coding work was done by SRC's Department of Survey Technologies,
and included check-coding at 5% overall with checking more heavily
concentrated on new coders and their initial cases.  Because of
this use of check-coding as further training for new coders, the
error rate was 12.7%.

Another reliability test compared independent coding of the same 50
cases by five experienced coders.  The 50 test cases were randomly
selected.  Rates of difference for this phase were 92% for
occupation and 86.5% for industry.

As stressed above in Sections I and II, not every Head's or Wife's
occupation was recoded.  The Head or Wife had to qualify under the
selection criteria--he or she had to be living as of 1992, an
original sample member, and had to report a minimum of three main
jobs between 1968 and 1992 with at least one report during the
1968-1980 period.  All original sample members who were known to be
deceased as of 1992 and who had at least one report between 1968
and 1980 were also included.

Thus, nonsample Heads and Wives, those who had married into the
study, were completely excluded.  Also excluded were those who had
never worked and those who were living in 1992 but had a total of
less than three reports since the beginning of the study.


Section IV: Linking Records

The Retrospective Occupation-Industry Files have an extremely
limited, albeit useful, amount of information, and the analyst will
want to access the much greater volume of data available for Heads
and Wives on the main Family and Individual Files.  The main files
can provide substantial information for many individuals dating
back to 1968, when the PSID began.  For others, though, the time
span of available data is more limited.


Using the 1968-1980 Retrospective Occupation-Industry Files with
the Cross-Year Individual Files

The 1968-1980 Retrospective Occupation-Industry Files can be linked
with any cross-year Individual File, since individual identifiers,
which can be corrected for subsequent releases of individual data
(although this is rare), are not included.

Data processing is required to link records between these two
files.  To accomplish linkages, match on yearly family identifiers.
Note that, as with other family files, the Retrospective Files have
variables specific to both Heads and Wives on each record.
Matching to individual data must take account of the individual's
relationship to Head as surely as matching any family file with
individual data.  Choose Head's occupation/industry variables for
an individual who is the current Head in the wave indicated;
similarly, select Wife's occupation/industry variables if the
person is Wife in that year.

                    MAIN         RETROSPECTIVE
                    INDIVIDUAL   OCCUPATION/
   WAVE OF DATA     FILE ID      INDUSTRY FILE ID
        1968        V30001            ID68
        1969        V30020            ID69
        1970        V30043            ID70
        1971        V30067            ID71
        1972        V30091            ID72
        1973        V30117            ID73
        1974        V30138            ID74
        1975        V30160            ID75
        1976        V30188            ID76
        1977        V30217            ID77
        1978        V30246            ID78
        1979        V30283            ID79
        1980        V30313            ID80

Note that not all Heads or Wives in a given wave have a match on
the corresponding Retrospective Occupation-Industry File.  As
mentioned above, only certain Heads and Wives were eligible for
recoding.


Using the 1968-1980 Retrospective Occupation-Industry Files with
the Family Files

Merging information from the single-year Family Files is quite
simple.  Match a Retrospective Occupation-Industry File with its
Family File using the yearly Interview (ID) Number.  The table
below lists corresponding variables from each file.

                    MAIN         RETROSPECTIVE
                    FAMILY       OCCUPATION/
   WAVE OF DATA     FILE ID      INDUSTRY FILE ID
        1968            V3            ID68
        1969          V442            ID69
        1970         V1102            ID70
        1971         V1802            ID71
        1972         V2402            ID72
        1973         V3002            ID73
        1974         V3402            ID74
        1975         V3802            ID75
        1976         V4302            ID76
        1977         V5202            ID77
        1978         V5702            ID78
        1979         V6302            ID79
        1980         V6902            ID80

Again, we cannot mention too often that not every Head or Wife has
a match on the related Retrospective Occupation-Industry File.
 



Institute for Social Research | University of Michigan | Privacy | Conditions of Use