🚀 go-pugleaf

The following are a bunch of replies to make posting regarding the
rumored problems with the 1990 PUMS.

-------------------------------------------------------------------------
Alexander Cavallo                      ST801840@BROWNVM.BROWN.EDU
Brown University                       alc@pstc3.pstc.brown.edu
Providence, RI  USA
-------------------------------------------------------------------------

(begin included messages)

Return-Path: <@BROWNVM.BROWN.EDU:ANDQC@CUNYVM.CUNY.EDU>
Received: from CUNYVM.CUNY.EDU (NJE origin MAILER@CUNYVMV2) by
          BROWNVM.BROWN.EDU (LMail V1.1d/1.7f) with BSMTP id 3509; Tue,
          3 Aug 1993 07:51:55 -0400
Received: from CUNYVM.CUNY.EDU (NJE origin ANDQC@CUNYVM) by CUNYVM.CUNY.EDU
          (LMail V1.1d/1.7f) with RFC822 id 2127; Tue, 3 Aug 1993 07:52:21 -0400
Date:         Tue, 03 Aug 93 07:49:00 EDT
From:         "Andrew A. Beveridge" <ANDQC@CUNYVM>
Subject:      Re: 1990 PUMS
To:           ST801840@BROWNVM

Dear Alex:

To scare the &%?@ out of you, here is the unedited material I have on the
PUMS problems.










======================================================================== 32
Return-Path: <@CUNYVM.CUNY.EDU:OR-L@UALTAVM.BITNET>
Received: from CUNYVM.BITNET (NJE origin LISTSERV@CUNYVM) by CUNYVM.CUNY.EDU
          (LMail V1.1d/1.7f) with BSMTP id 7620; Tue, 6 Apr 1993 14:15:04 -0400
Date:         Tue, 6 Apr 1993 14:03:37 EDT
Reply-To:     "An Informal List for Official Representatives of the ICPSR"
              <OR-L@UALTAVM.BITNET>
Sender:       "An Informal List for Official Representatives of the ICPSR"
              <OR-L@UALTAVM.BITNET>
From:         Judith Rowe <JUDITH@PUCC.BITNET>
Subject:      Re: access to census data
To:           Andrew Beveridge <ANDQC@CUNYVM.BITNET>
In-Reply-To:  Message of Mon,
              5 Apr 1993 10:02:17 EST from <lynn@EROS.CC.COLUMBIA.EDU>

Rumors on the PUMS problem are now rampant.  May I try to clarify some of
the problems.

1) It is likely that the four additional New England states will be added
to the list of 14 states already distributed.  I can find no one at Census
whose list includes Maryland as a problem state.

2) Within two weeks Census expects to release a final list of states and PUMAS
with problems.
At that point sales and distribution will resume on all OTHER
states, i.e. all states without problems. ICPSR has withheld distribution of
PUMS at the request of the Census Bureau.  Once Census resumes distribution
ICPSR should be able to follow suit.

3) If a state has even one PUMA with a problem the entire state will be
released.  Replacement tapes will be forthcoming in the early summer.
z OR-L UALTAVM  4/06/93
'Judith Rowe         Andrew Beveridge     4/06/93*access to census data
======================================================================== 92
Return-Path: <@CUNYVM.CUNY.EDU:OR-L@UALTAVM.BITNET>
Received: from CUNYVM.BITNET (NJE origin LISTSERV@CUNYVM) by CUNYVM.CUNY.EDU
          (LMail V1.1d/1.7f) with BSMTP id 5666; Tue, 6 Apr 1993 16:23:28 -0400
Date:         Tue, 6 Apr 1993 15:55:38 -0400
Reply-To:     "An Informal List for Official Representatives of the ICPSR"
              <OR-L@UALTAVM.BITNET>
Sender:       "An Informal List for Official Representatives of the ICPSR"
              <OR-L@UALTAVM.BITNET>
From:         Peter Joftis <pmj@TDIS.ICPSR.UMICH.EDU>
Subject:      ICPSR Statement on PUMS
X-To:         or-l@ualtavm.bitnet
To:           Andrew Beveridge <ANDQC@CUNYVM.BITNET>

                                                    April 6, 1993

ICPSR staff has been closely following the messages in this list
server about the PUMS problems, and has been trying to gather
information from elsewhere about this situation. It is clear that the
problems with the PUMS series are vexing the research community a
great deal. Here's what ICPSR has been able to discover, and what
we're doing about it.

The Census Bureau discovered some (presumably major) problems with the
assignment of respondents to sub-state Public Use Microdata Areas
(PUMAs).  Until they could resolve those problems, the Bureau put a
hold on further distribution of both the 1% and 5% files for the whole
nation. ICPSR followed the Bureau's lead in temporarily ceasing
distribution of these files until more could be learned about the
nature and extent of the problems. Now other information is trickling
in, both from Bureau investigations of the problem and from
researchers who are beginning to report odd things that they have
found in using the data. These range from coding anomalies to problems
with the weight variables, and include discrepancies between data
files and technical documentation.

As more information begins to surface, there is strong suspicion that
the problems with PUMS are more widespread than first announced by the
Bureau.  A thorough review of what is actually known about PUMS
problems was conducted by ICPSR's 1990 Census Advisory Committee at
its April 2, 1993 meeting in Detroit. After hearing of known and
suspected problems, the committee unanimously recommended that the
ICPSR "freeze" on PUMS distribution be continued until the Census
Bureau completes its investigation and decides upon a course of
corrective action.

A number of rumors about the PUMS problems have gained wide
circulation. Most of these are unfounded, only partly true, or
prematurely derived from scanty or incomplete information. Information
digested by the ICPSR Census Advisory Committee led to the following
conclusions:

   1) It is highly probable that errors in the PUMS files are more
      extensive than initially thought, affecting more than the
      sub-state PUMA location of respondents. Particular variables
      with odd distributional properties are being examined by the
      Bureau, and may have to be recreated for ALL respondents.

   2) There is no certainty about the specific states or areas of
      states that are affected by errors in these files. Lists of
      states with and without problems are in constant flux at this
      time.

   3) The nature of the problems known by and/or reported to the
      Census Bureau suggests that these data are not "safe" to use
      for substantive applications at the present time.

   4) In all likelihood, errors in these files will be handled by
      re-issuing of ENTIRE FILES, either for some states or for all
      states. The Census Bureau has no plans to re-issue parts of files.

The Advisory Committee is working with the Census Bureau to help
pinpoint some of the anomalies that have been reported. The
Committee's findings will be reported to the Bureau, and we expect to
receive timely notification from the Bureau on the results of their
investigations. The current status of PUMS and what ICPSR knows about
the problems will be posted in CDNet and in Series-L as developments
occur. (In addition, a copy of this notice will be sent to all ICPSR
Official Representatives.)

This situation presents major problems for the research community,
beset with deadlines for dissertations, publications and other
projects that require use of the PUMS files. It might be tempting to
minimize or ignore the extent of problems in these files, and press
ahead with research use of PUMS. ICPSR, its 1990 Census Advisory
Committee, and the Census Bureau all believe that it is unwise to do
so at the present time. When the integrity of these data files has
been verified (or the files resupplied), ICPSR will resume
distribution of the PUMS series to the academic community.

ICPSR Staff
z OR-L UALTAVM  4/06/93
'Peter Joftis        Andrew Beveridge     4/06/93 ICPSR Statement on PUMS
======================================================================== 14
Date:         Fri, 09 Jul 93 15:59:31 EDT
From:         "Andrew A. Beveridge" <ANDQC@CUNYVM>
Subject:      Re: CENSUS INQUIRIES
To:           Peter Granda <peterg@icpsr.umich.edu>
In-Reply-To:  Your message of Fri, 9 Jul 93 08:30:04 -0400

Dear Peter:

If you could FAX the page you refer to, I would be most appreciative.

Andy

P.S.  It is hard to keep track of the Census Bureau.
'Andrew A. Beveridge Peter Granda         7/09/93*CENSUS INQUIRIES
======================================================================== 156
Return-Path: <@CUNYVM.CUNY.EDU:peterg@ICPSR.UMICH.EDU>
Received: from CUNYVM (NJE origin SMTP@CUNYVM) by CUNYVM.CUNY.EDU (LMail
          V1.1d/1.7f) with BSMTP id 6738; Fri, 9 Jul 1993 08:29:50 -0400
Received: from icpsr.umich.edu by CUNYVM.CUNY.EDU (IBM VM SMTP V2R2) with TCP;
   Fri, 09 Jul 93 08:29:48 EDT
Received: by icpsr.umich.edu (5.65/icpsr-1.1)
        id AA00828; Fri, 9 Jul 93 08:30:04 -0400
Date: Fri, 9 Jul 93 08:30:04 -0400
From: Peter Granda <peterg@icpsr.umich.edu>
Message-Id: <9307091230.AA00828@icpsr.umich.edu>
To: ANDQC@CUNYVM.CUNY.EDU
Subject: CENSUS INQUIRIES

  Peter Joftis passed along your inquiry about PUMS and STF3 to the
  archive
section and we have the following to report:

  1)  We have noted a number of differences between the first and
  second
      releases of PUMS and have also heard from other users as well:

ICPSR compared the original and resupplied 1% PUMS for Delaware and
Maine.
For Maine, 81 variables had unequal values across the original and
resupplied
data - only two of these were PUMA variables.  The comparison of
Delaware
data revealed five variables with some observations unequal: PSA,
MIGSTATE,
MIGPUMA, POWSTATE, POWPUMA.

There are several typos in the original October, 1992 edition of
the PUMS codebook which were corrected in the January, 1993 edition.
These
included errors in the HISPANIC coding scheme.

Below is a message forwarded to ICPSR by CIESIN regarding errors in the
January, 1992 edition of the codebook. ICPSR is awaiting confirmation
of these
errors by the Census Bureau.

>From rich.robinson@qm.ciesin.org Thu Jul  1 09:44:20 1993
Received: from netnews.ciesin.org by icpsr.umich.edu (5.65/icpsr-1.1)
        id AA03615; Thu, 1 Jul 93 09:44:19 -0400
Received: from qm.ciesin.org by mail.ciesin.org with SMTP id AA07072
  (5.65c/IDA-1.4.4 for <alon@tdis.icpsr.umich.edu>); Thu, 1 Jul 1993
  09:44:18 -0400
Message-Id: <199307011344.AA07072@mail.ciesin.org>
Date: 1 Jul 1993 09:45:47 U
From: "Rich Robinson" <rich.robinson@qm.ciesin.org>
Subject: PUMS/Explore Account
To: "Alon Axelrod" <alon@tdis.icpsr.umich.edu>
Status: R

< stuff deleted here >

The following message was sent to us by Erin Kehoe at the University of
Wisconsin. Please make note of the code errors he has identified.

I spoke yesterday (6/14) with Romala Basu at the Census Bureau about 3
variables
on the person record which we were having difficulty understanding.  It
turns
out 2 of them have the codes reversed in the codebook, while the other
is just
unclear.  Here is her explanation:

--------------------------------------------------------------------------------
ROWNCHLD (P32)  SHOULD BE:
        1 = Own Child
        0 = Not Own Child

(these are reversed on page 5-19 of the codebook)
--------------------------------------------------------------------------------
RRELCHLD (P34)  SHOULD BE:
        1 = Related Child
        0 = Not Related Child

(again, these are reversed on page 5-19)

Another note about rrelchld -- the housing record has a variable with
the exact
same name, so one should be changed slightly in any dictionary being
used to
avoid strange results.
--------------------------------------------------------------------------------
RAGECHLD (P33)  is Presence and age of own children FOR FEMALES -- this
last
part about females is not made clear.  The values, which are also
unclear when
you start working with them, are as follows:

0 = N/A ( males ; OR females under 16 years old)
1 = FEMALES with own children under 6 years only
2 = FEMALES with own children 6-17 years only
3 = FEMALES with own children under 6 years and 6-17 years
4 = FEMALES 16 YEARS OR OVER with no own children
--------------------------------------------------------------------------------

Romala said that the variables with an extra 'R' on the front such as
these 3
are recodes that were created to help users avoid complex select
statements
(this is explained somewhere in the codebook).  Unfortunately, not only
is
the codebook wrong on some and confusing on others, but some do not
have
enough values.  Two examples are ROWNCHLD and RRELCHLD, above.  They
should
both have an N/A category.

The Bureau is suggesting users avoid using these 'recode' variables if
they do
not give feasible results.  Each of them can be obtained by using a
combination
of other variables in the file, it's just the select statements will
become
more complex.

Please pass this information along to others who may be using 1990 PUMS
data.
We will watch for other errors that may surface.


-Erin Kehoe
 Applied Population Lab
 Dept. of Rural Soc.
 UW-Madison, WI
 6/15/93

In terms of resupplied PUMS data, we have just received the
updated 1 percent files but are still waiting
to hear if there will be new documentation or not.  5 percent files
have
been ordered but we still have not received any as yet.


2)  There are more differences in STF3 than just in the disability
variables.  We have not received any user or technical note which
describes these differences in full detail but there is an explanation
of sorts (which you may have already seen) in the September, 1992
number of CENSUS AND YOU.  Page 4 describes briefly problems with:
MOBILITY LIMITATION, SELF-CARE LIMITATION, WEEKS WORKED IN 1989,
USUAL HOURS WORKED PER WEEK WORKED IN 1989, NEGATIVE INCOME, AND
PUBLIC ASSISTANCE.  I could fax you this page if you do not have a
copy around already.


I hope this is of some help to you.
                                     Peter Granda
                                     ICPSR



z peterg ICPSR  7/09/93
'Peter Granda        ANDQC@CUNYVM.CUNY.E  7/09/93 CENSUS INQUIRIES
======================================================================== 14
Date:         Fri, 09 Jul 93 16:02:32 EDT
From:         "Andrew A. Beveridge" <ANDQC@CUNYVM>
Subject:      Re: CENSUS INQUIRIES
To:           Peter Granda <peterg@icpsr.umich.edu>
In-Reply-To:  Your message of Fri, 9 Jul 93 08:30:04 -0400

Dear Peter:

The FAX number is 718-520-7241.

Sorry.

Andy Beveridge
'Andrew A. Beveridge Peter Granda         7/09/93*CENSUS INQUIRIES
======================================================================== 12
Date:         Sat, 10 Jul 93 09:55:34 EDT
From:         "Andrew A. Beveridge" <ANDQC@CUNYVM>
Subject:      Re: CENSUS INQUIRIES
To:           Peter Granda <peterg@icpsr.umich.edu>
In-Reply-To:  Your message of Fri, 9 Jul 93 16:52:18 -0400

Dear Peter:

I will check it out on Monday.  Sorry.

Andy Beveridge
'Andrew A. Beveridge Peter Granda         7/10/93*CENSUS INQUIRIES
======================================================================== 18
Date:         Tue, 13 Jul 93 10:17:53 EDT
From:         "Andrew A. Beveridge" <ANDQC@CUNYVM>
Subject:      FAX, Census and Power Outage
To:           Peter Granda <peterg@icpsr.umich.edu>

Dear Peter:

Our FAX was out due to a power failure brought on by the Heat Wave
here.  Its working and its number is 718-520-7241.

Please send me the information from Census and You.

Thanks a lot.

It will help us in NYC.

Andy
Return-Path: <@BROWNVM.BROWN.EDU,@YALEVM.CIS.YALE.EDU:SSDA@YALEVM.CIS.YALE.EDU>
Received: from YALEVM.CIS.YALE.EDU (NJE origin MAILER@YALEVM) by
          BROWNVM.BROWN.EDU (LMail V1.1d/1.7f) with BSMTP id 4531; Tue,
          3 Aug 1993 08:19:25 -0400
Received: from YALEVM (NJE origin SSDA@YALEVM) by YALEVM.CIS.YALE.EDU (LMail
          V1.1d/1.7f) with BSMTP id 7923; Tue, 3 Aug 1993 08:19:14 -0400
Date:         Tue, 03 Aug 93 08:16:28 EDT
From:         JoAnn Dionne <SSDA@YALEVM>
Subject:      Re: 1990 PUMS
To:           Alex Cavallo <ST801840@BROWNVM>
In-Reply-To:  Your message of Thu, 29 Jul 1993 20:28:07 EDT

Listed below is a copy of a fax send out by the Census Bureau in March.  It
is my understanding that the corrected 1990 PUMS are now being released.
At the end of the fax message is an addenda from the University of Wisconsin
on weights.  I hope this answers your questions--but if not please get back
in touch.

                                            JoAnn Dionne

Received via FAX 23-MAR-1993

United States Department of Commerce
Bureau of the Census
Washington DC 20233-0001

Census of Population and Housing 1990:
Public Use Microdata Samples - United States

ALERT

There is a problem with the 5% and 1% PUMS files.  The problem affects certain
sub-county level PUMAs (Public Use Microdata Areas.)  Customers using PUMS
at the state or higher levels of geography are not affected by the problem;
however users of county or sub-county level PUMAs may have a problem.

We are determining now the number of files affected.  To correct the problem,
we may have to rerun (reissue) several files, both 1% and 5%.

After we have rerun and reviewed the files, we will issue replacement tapes for
all states affected.  Until that time, we will suspend PUMS sales for all
states.

We project it may take as long as two months before replacement files will
be issued.  However, we will reissue the affected files as soon as it is
possible.  This, also, will delay the release of the PUMS CD-ROM and
any subsequent PUMS files that uses sub-state geography (elderly, Puerto
Rico).  The other outlying areas PUMS files have no geography other than
"state".

--=-=-=-=-=-= end of Census Bureau Alert -=-=-=-=-=-=-=-=-=

Another note on PUMS90:
It has been found at CDE that there are 0 person weights in a few cases.  When
asked, people at the Bureau said that, since the file has integer weights,
any decimal weights generated in processing that were less than .5 rounded
down to 0.  The documentation for PWGT1 (person record cols. 18-21) should
show a range of 0 to 1152 rather than the range of 1 to 1152 printed in
the current version.  This will be corrected in a user note.

                                                        (ALC - 23-MAR-1993)