Docstoc

2008 ASEC (Word format) - The TRIM3 system converts raw

Document Sample
2008 ASEC (Word format) - The TRIM3 system converts raw Powered By Docstoc
					                    THE URBAN INSTITUTE
                                       MEMORANDUM

TO:             Interested Persons
FROM:           Joyce Morton
DATE:           November 5, 2009
SUBJECT:        Conversion of the 2008 ASEC


        This memorandum describes the processing of the public-use 2008 Current Population
Survey (CPS) Annual Social and Economic Supplement (ASEC) so the data can be used in
TRIM3 simulations. The 2008 ASEC provides TRIM3 input data for calendar year 2007
simulations. The final CY 2007 TRIM3 input data include virtually all the information in the
public-use ASEC data as well as other information needed for TRIM3 simulations including
monthly income variables, edited and recoded variables expected by the TRIM3 code, random
numbers needed for various imputations and behavioral decisions, and imputed legal status for
non-citizens.

        The 2008 ASEC data are largely the same as for the previous year with no major changes
to the survey or the data, and procedures for converting the ASEC data to a form usable by
TRIM3 were also largely identical to procedures used for the conversion of 2007 survey data.
One technical change is described in the section on replicating households.

          This document includes information on the following topics in the pages following.
      -      Tasks Performed by Conversion
      -      2008 ASEC Record Counts and Weighted Totals
      -      Appendices
            - Appendix A, Comparison of Wt. Mean Income Amounts on the 2008 and 2007
                ASEC
            -   Appendix B, Number of Households and Persons by State 2008 ASEC/TRIM3 CY2007
            -   Appendix C, Variables showing the Greatest Change in Weighted Sum of Values

                                 Tasks Performed by Conversion

        TRIM3 software converts raw survey data to standard TRIM3 format so that the data may
be used for microsimulations. The conversion procedure recodes any revised ASEC codes to
match existing ones already in use in TRIM3, creates new variables needed by the
microsimulation model, creates monthly values from some annual amounts, creates an inflation
index used in TRIM3 simulations, generates random numbers required by simulations, and loads
the data into TRIM3 database tables.

        Conversion makes it possible for a survey file to be used in TRIM3 without making any
changes to the model's computer code. Two different surveys could use the same sampling units
and collect the same variables while using completely different coding schemes for those
variables and a different ordering of variables on input records. Without conversion, there would
be a need for different versions of the TRIM3 model to read the variables from different
locations and interpret the codes in different ways.
        In practice, no two surveys collect exactly the same information. Similar information
may be available but collected in different ways. For example, information contained in a
variable on one survey may have to be pieced together from several variables on another survey.
The ASEC itself changes each year; new variables are added, others are dropped, and coding
schemes are changed, and different methods are used to create the same variables. Generally,
these changes are minor, but periodically there are major changes in the processing and editing
methods used by the Census Bureau. These changes are all handled by the TRIM3 conversion
process.

       Conversion has both technical and substantive features. From a technical viewpoint,
conversion reformats data so that it may be efficiently processed by existing simulation code.
Substantive aspects of conversion include:
       -   editing the data
       -   recoding variables and creating new ones
       -   allocating income across the months of the year
       -   generating random numbers
       -   replicating households

Reformatting the Data

        The ASEC public-use file includes household, family, and person records. During
conversion of a file, ASEC data are stored in the TRIM3 database in household, family, adult,
and person database tables. Adult tables contain variables that pertain to "economic adults,” i.e.,
persons aged 15 and older. Person tables contain variables coded for all persons regardless of
age. Monthly variables created during the conversion are stored in a separate table in the
database which has 12 monthly records for each adult. The database design allows information
to be stored in an efficient manner where it can be easily accessed when processing households.

        Empty (non-interview) ASEC households are excluded from the TRIM3 database.
Conversion also checks for and excludes households consisting entirely of children under the age
of 15, though in recent years no child-only households have been found. If a child under the age
of 15 has no relatives in a household, TRIM3 places the child in the family of the household
reference person, the primary family. Because child-only families are merged into primary
families, there are somewhat fewer TRIM3 families than there are ASEC families in the final
dataset.

Editing the Data

        Although ASEC files are generally well edited by the Census Bureau, a variety of
additional edits are performed as part of the conversion. In general, conversion procedures
confirm that variable codes and universes are as specified in a survey's codebook. The
conversion code flags out-of-range and unexpected values so that potential problems can be
identified and corrected before the data are brought into the TRIM3 database.

         We also generate post-conversion term statistics that are compared with the same
statistics generated from the previous years’ data to identify variables showing significant
changes. We investigate any variables that show unexpected changes and make any corrections
or adjustments in conversion procedures that are required or simply document the changes in the
TRIM3 dictionary.
Recoding Variables and Creating New Ones

       Even when the same information is collected by two surveys, different coding schemes
may be used. When required, conversion software recodes variables to have the coding schemes
expected by the TRIM3 simulation modules. For instance, the TRIM3 variable
HighestGradeCompleted, which has a range of 0 to 18, is a recode of information contained in
an ASEC field for educational attainment that has codes of 0 and 31 to 46. With this recoding,
TRIM3 simulation code is able to treat HighestGradeCompleted the same from one survey year
to another and from one survey source to another.

        Conversion also creates new variables from existing ones in order to increase processing
efficiency. For example, broad income measures are generated by summing detailed income
components; new health insurance variables are generated by combining initial reported
information with information from final "catch all" questions; and standard TRIM3 "family type"
variables widely used throughout TRIM3 are constructed. If a raw survey file does not have a
variable needed by TRIM3, the variable may be coded from other information that is available
during conversion.

Allocating Income Across Months of the Year

        The ASEC and most of the other surveys that might be used by TRIM3 report annual
income amounts. However, government transfer programs calculate monthly eligibility and
benefits using monthly income information. The conversion procedure must therefore allocate
annual income amounts to months of the year.

        Monthly allocation uses ASEC-reported data on number of weeks of employment and
number of different spells of work for the calendar year preceding the March survey. The
different spells are distributed over the 52 weeks of the year, and the weekly information is then
summarized by monthly variables.

        Beginning with the conversion of 2003 ASEC survey data, all months are assigned an
equal number of weeks (4.33333).1 Even before this change, some unearned income amounts
were assigned evenly over the months. And even with this change, equal monthly earnings
across all months is not guaranteed. However, with the change to 4.3333-week months, if a
person worked full-time all 52 weeks of the year, his/her monthly earnings will all be equal.

       Prior to converting ASEC data, the appropriate annual series of monthly employment and
unemployment totals are made available to conversion software as targets, and with that
information, weeks of work and unemployment are assigned to the various months so as to
approach those monthly targets. Once periods of employment and unemployment have been
assigned to each adult on the basis of his or her reported information, wages, self-employment
income, and farm income are allocated over the year consistent with assigned weeks of work and
unemployment. Unemployment compensation is generally divided over weeks of
unemployment, but for a portion of recipients a one-month lag in receipt of benefits is simulated.

       Workers' compensation is generally divided over all weeks in which a person was either
unemployed or out of the labor force; but a random subset of recipients may be simulated to
receive their workers' compensation as a lump sum, all in one month.
1
 Prior to that time, four months were assigned 5 weeks, and the remaining months were assigned 4 weeks. Weeks
with days in two different months were assigned to the month in which most of its days fell.
        The number of months over which alimony and child support income is allocated is
determined probabilistically on the basis of look-up tables generated from SIPP data. Beginning
with the 2001 ASEC conversion, separate look-up tables were made available for TANF
recipients and non-TANF recipients using data from the 1996 SIPP on alimony and child support
recipiency patterns of these two groups. The tables provide the proportion of persons receiving
alimony or child support income in differing numbers of months by amount of income received.
As an example, SIPP data may show that of the persons who have either alimony or child
support income of $5,000 or more a year and no TANF income, 80 percent receive that income
during all 12 months of the year. On the other hand, SIPP data may show that persons who have
either alimony or child support income of between $1,000 and $2,000 a year who also receive
TANF have just a 24 percent liklihood of receiving the alimony or child support income during
all 12 months of the year, an 8 percent probability of having received the income for 11 months,
4 percent probability of having received the income in just 10 months, etc. These probabilities
are used to allocate reported income amounts over the likely number of months the income was
received.

      Persons who receive both Child Support and/or Alimony and TANF and whose Child
Support plus Alimony divided by months of TANF receipt is equal to their state's pass-through
amount are excluded from the look-up table. Rather, their months of Alimony and/or Child
Support receipt is set equal to the reported number of months of TANF receipt.

         All other types of income are divided evenly over the 12 months of the year. Income
that is evenly distributed over the year includes asset income (combined income from rents,
royalties, interest, dividends, and estates or trusts) and the Social Security, Railroad Retirement,
pension income, veterans payments, regular contributions, and other income components of
TRIM3’s MonthlyUnearnedIncome field.2

Generating Random Numbers

        Random numbers are used in many TRIM3 modules to determine whether a person or
unit with a certain probability of some outcome will actually have that outcome. For example,
random numbers are used in deciding whether a unit that is eligible for TANF will actually
receive TANF benefits. A unit's characteristics determine its probability of participation, and a
random number between 0 and 1 is compared to that probability to determine whether the unit
will participate. If a probability is .73, any random number of .73 or less results in participation.
Random numbers are also required by various imputation functions used when variables needed
by the model are not present on the raw survey.

        Some TRIM3 simulations create random numbers when those simulations are run; other
random numbers are generated during conversion and are stored in database tables. At whatever
point they are generated, TRIM3 generates different random numbers for each task requiring
them so that there will be no unintended relationship between, for example, the Food Stamp
Program participation decision and imputed child care expenditure amounts. The random
number seed used in generating a person's random number is derived from the name of the
random number variable, the data year, household or person identifier, and month (if a monthly
random number). This ensures that if a module using a random number is run twice on the same
input file with the same program rules, the results will be exactly the same.

2
 The two other components of the MonthlyUnearnedIncome field, Alimony and Child support, are divided as
explained in the preceeding text.
Replicating Households

        We perform two separate conversions of ASEC data. We initially convert ASEC data
without any household replication. Then, once imputed immigrant status information is
available, we merge that information into ASEC data, replicating certain immigrant households
in the process. When replicating households, we split household, family and person weights so
that we can match population subgroup targets and yet weight totals remain the same. Beginning
with the 2004 ASEC conversion, we added a second household replication and split of weights
so that when TRIM3 data are matched with Statistics of Income (SOI) data, there are more
unweighted high-income households for matching with SOI high-income units.

         Details of the Immigrant Replication. The legal status imputation process determines
whether a household with immigrants is replicated, the number of replications, and the portion of
a household's weight to assign to each replicate. For example, for a two-person household with
one non-citizen, if the non-citizen was determined to have a 40 percent probability of being an
undocumented alien and a 60 percent chance of being a legal permanent resident (LPR), the final
data would include two versions of the household. The first version would have 40 percent of
the weight and the non-citizen would be coded as undocumented, while the second version
would have 60 percent of the weight and the non-citizen would be coded as an LPR. Larger
households might have additional replicates. Some immigrant households may not be replicated
at all; for instance, if all non-citizens in a household are determined to be refugees, no replication
would be needed. The maximum number of replicates for any given household in the CY 2007
data is six, including the original instance of the household.

        Details of the SOI High-income Split. To increase the unweighted count of high-income
households in TRIM3 data, we identify high-income cloning candidates and create four
additional copies of the selected households. Thus, each high-income clone is represented in the
TRIM3 data five times, and household, family, and person weights are equally split among the
clones so that they sum to the original ASEC weights.

        To be identified as a high-income cloning candidate, a household must consist entirely of
U.S. natives to avoid the possibility of replicating the same household for both the high-income
and immigrant-imputation cloning. In addition, households are required to have a family with a
PovertyRate at least 4.5 times the family’s poverty threshold; and the household must have at
least one person with a topcode earnings flag indicating earnings were topcoded (ASEC fields
tcernval, tcwsval, tcseval, or tcffmval), or a top-coded value for one of the following income
types: survivors income (sur-val1 or sur-val2), retirement income (ret-val1 or ret-val2), interest
income (int-val), dividend income (div-val), rental income (rnt-val), alimony income (alm-val),
or other income (oi-val).

        We changed the coding of HighIncomeClone beginning with the conversion of the
default CY 2005 data. In prior years, this variable is a flag that is coded either 1 for households
selected for high-income cloning or 0 for all other households. Beginning with the default 2005
data, HighIncomeClone is coded from 1 to 5 for households selected for cloning so that a single
clone may be identified in work for which this information is needed. Most households are
coded 0, which indicates they are not high-income clones.
                       2008 ASEC Record Counts and Weighted Totals

        The following tabulations show pre-conversion unweighted counts and post-conversion
weighted and unweighted counts of households, families and persons from the 2008 ASEC
compared with similar post-conversion counts from the 2007 ASEC. Table A provides counts
in the unadjusted (alternative) TRIM3 data. In these data, original pre-conversion unweighted
record counts include empty (non-interview) households (21,630) and child-only families (256)
that conversion merged with primary families or single household heads. Without the empty
households and with the child-only families merged with other families or household heads,
there are 75,872 unweighted households in the data—very close to the same number of
households as were interviewed the preceeding year.

                                Table A.Unadjusted Alternate Data

                     Pre-conversion         Post-conversion              Post-conversion
                      2008 ASEC               2008 ASEC                    2007 ASEC
                      Unweighted       Unweighted      Weighted    Unweighted      Weighted
     Households               97,502         75,872    116,880,816      75,447       116,131,345
     Families                 86,886         86,630    134,121,050      85,905       132,737,301
     Persons                 206,404       206,404     299,105,719     206,639       296,824,002
     Adults                  156,633       156,633     238,147,510     155,954       236,019,650


        Table B provides record counts in the immigrant- and high income-adjusted (default) data
that contain replicate households. After replications, there are 93,657 unweighted households in
the CY 2007 data. For data in which there is a very small difference in weighted number of
households in the alternative and default schemas, the difference is due to different householders
being identified by Census procedures than were identified by the TRIM3 convert procedures in
a very small number of households, typically no more than five. Since household weight is
always set to the person weight of the householder, this can cause a slight difference in the total
weighted household counts. Census appears to have recently changed their procedures so that
these minor differences no longer occur. For CY 2007, the weighted household totals match
exactly.

                   Table B. Immigrant- and High-income Adjusted Default Data
                     Pre-conversion         Post-conversion               Post-conversion
                      2008 ASEC               2008 ASEC                     2007 ASEC
                      Unweighted       Unweighted      Weighted     Unweighted      Weighted
      Households              97,502        93,657    116,880,816       92,520       116,131,347
      Families                86,886      107,537     134,121,050      105,871       132,737,304
      Persons               206,404       260,189     299,105,721      258,684       296,824,007
      Adults                156,633       197,324     238,147,510      195,073       236,019,653
                                           Appendices

        The appendices that follow provide some detailed information that has historically been
of interest to analysts for an initial assessment of changes in sampling distribution and data
trends and for spotting unexpected statistics that may warrant further investigation.

        Appendix A compares mean income amounts in the 2008 ASEC (CY 2007) with those in
the 2007 ASEC (CY 2006). The variables that are tabulated provide analysts with some key
measures for assessing changes in survey data from one year to the next. The table includes
columns that show both the number of persons receiving a particular type of income, the total
amount received, and the average amount received for both years. The final three columns show
the percent by which each of these statistics (number of persons, total amount received, and
average amount received) either increased or decreased from the earlier to the latter year.

        Many income types show little change from one year to the next. For instance, the
average amount of wage and salary income (TotalWages) increased by 1.87 percent from CY
2006 to CY 2007, and the number of persons with wage or salary income increased by just under
1 percent. Since both of these measures went in the same direction, the combined impact can be
seen in the 2.88 percent increase in total amount of wage and salary income received in CY 2007
relative to CY 2006. The situation shown by the Interest variable is somewhat different in that
number of persons receiving the income decreased (by 1.7 percent), but the average amount of
income received increased (by 4.3 percent). Since those measures went in opposite directions,
they had a mixed impact on the total amount of interest reported, which increased by 2.54
percent from the earlier year to the latter one. Typically, incomes that are reported by the largest
number of persons—e.g., wage and salary income and interest income—are more stable from
year to year than incomes that are reported by relatively few persons—e.g., farm self-
employment and the combined rental and royalty income measure.

       Appendix B shows the number of households and persons by state. The unadjusted
unweighted number of households and persons are counts from the public use data following
conversion but without any replication of households. The adjusted counts include replicate
households created for the immigrant imputation and other replicate households to increase to
the number of high income households. For households that have been replicated, the original
household, family, and person weights have been split so that total weights sum to the original
ASEC weights. Weights in households replicated to increase the number of high income
households are evenly split among the replicates, whereas weights in households replicated for
the immigrant imputation are split to match citizenship targets.

        Appendix C shows weighted statistics for some of the variables showing the most change
from CY 2006 to CY 2007, as measured by the weighted sum of values. The table separates
allocation flags—which are shown in the bottom half of the table—from other variables. Within
those broad classifications, variables are ordered by the ones showing the largest negative change
to those showing the largest positive change. Usually, a negative change means that there are
either significantly fewer people being coded with a non-zero value, or there is a significant
decrease in the average value of the variable for persons who have a non-zero value, or both.
Sometimes, a very large change in frequency will offset a change in magnitude, or the reverse
will occur—i.e., a change in magnitude will offset a change in frequency. Positive changes, of
course, go in the opposite direction.
                                                            Appendix A
                                      Comparison of Wt. Mean Income Amounts
                                           On the 2008 and 2007 ASEC

                                2008 ASEC (CY 2007)                      2007 ASEC (CY 2006)                  Percent Difference
                           Persons      Total     Average           Persons      Total     Average         Number Total        Avg
                          Receiving    Received  Received          Receiving    Received  Received         Persons    Amt     Amt
                                        (millions)                                (millions)
                                                            Person-level
TANF (formerly             1,228,122        $3,990      $3,248     1,293,090         $4,394      $3,398      -5.0%    -9.2%      -4.4%
AFDC)
Alimony                      411,218       $5,461      $13,281        394,625        $4,657     $11,801       4.2%    17.3%      12.5%
BusinessSelf              12,499,066     $396,048      $31,686     13,099,400      $411,670     $31,427      -4.6%    -3.8%       0.8%
Employment
ChildSupport               4,963,881      $24,663       $4,968      5,132,380       $25,839      $5,034      -3.3%    -4.6%      -1.3%
DetailedOtherIncome        8,542,340      $48,620       $5,692      8,067,630       $47,003      $5,826       5.9%     3.4%      -2.3%
DividendsEstates          33,647,759     $128,328       $3,814     33,405,800      $122,598      $3,670       0.7%     4.7%       3.9%
OrTrusts
FarmSelfEmployment         2,048,102      $30,228      $14,759      2,326,860       $31,382     $13,487     -12.0%    -3.7%       9.4%
GeneralOtherIncome        15,027,695      $92,714       $6,170     14,793,700       $92,295      $6,239       1.6%     0.5%      -1.1%
GovernmentPensions         6,781,830     $156,168      $23,027      6,988,800      $148,470     $21,244      -3.0%     5.2%       8.4%
Interest                  90,997,941     $243,084       $2,671     92,574,800      $237,062      $2,561      -1.7%     2.5%       4.3%
OtherPublic                  637,353       $2,184       $3,427        618,860        $1,839      $2,972       3.0%    18.8%      15.3%
Assistance1
OtherRegular               2,035,932       $13,969      $6,861      2,161,040       $14,796      $6,847      -5.8%    -5.6%      0.2%
Contributions
PrivatePensions           12,279,277     $160,759      $13,092     12,269,800      $158,411     $12,911       0.1%     1.5%      1.4%
PublicAssistance2          6,642,120      $39,182       $5,899      6,637,440       $38,210      $5,757       0.1%     2.5%      2.5%
RentsOrRoyalties          8,378,939        $68,184      $8,137     10,337,200       $70,680      $6,837     -18.9%    -3.5%      19.0%
Social SecurityOrRrr     42,079,248      $497,215      $11,816     41,377,100     $474,677      $11,472       1.7%     4.7%       3.0%
SSI                       5,039,295        $33,008      $6,550      4,991,880       $31,977      $6,406       0.9%     3.2%       2.3%
TotalWages              149,437,198     $6,135,611     $41,058    147,971,000    $5,963,714     $40,303       1.0%     2.9%       1.9%
Unemployment              5,200,329        $21,876      $4,207      5,230,010       $20,666      $3,952      -0.6%     5.9%       6.5%
Compensation
VeteransPayments           2,533,438       $29,092     $11,483      2,416,450       $27,605     $11,424       4.8%     5.4%       0.5%
Workers                    2,065,977       $22,839     $11,055      2,105,030       $24,074     $11,437      -1.9%    -5.1%      -3.3%
Compensation
                                                         Household-level
                         Households       Total       Average  Households          Total       Average      Hhlds     Total      Avg
                         Receiving        Value        Value    Receiving          Value        Value      Recving    Value      Value
                                        (millions)                                (millions)
FoodStampsValue            7,628,166     $16,905       $2,216       7,281,969       $15,878      $2,180       4.8%     6.5%      1.7%

1
 OtherPublicAssistance is LastOtherAllocatedPublicAssist + LastOtherIncome (if the source field is "Other Public Assistance").

2
 PublicAssistance is a computed field. It is composed of SSI, TANF, and OtherPublicAssistance. SSI is taken directly from the ASEC.
TANF consists of LastOtherIncome (read directly from the ASEC) (if the associated field indicating type of income received is coded
AFDC) + LastAllocated AFDC. LastAllocatedAFDC is allocated from LastPublicAssistanceAmount (read directly from the ASEC) if
the source field is coded AFDC/TANF or both AFDC/TANF and other PA. TANF is also reported separately in the AFDC field.
                                              Appendix B
                               Number of Households and Persons by State
                                     2008 ASEC/TRIM3 CY 2007

                                Households                                        Persons
                                                        Wt                                                Wt
State                            Adj %                 Adj %                      Adj %                  Adj %
FIPS            Unadj    Adj       of     Wt Adj         of     Unadj     Adj       of      Wt Adj         of
Code    State   Num1    Num2     Total    Number       Total    Num1     Num2     Total     Number       Total
   1    AL        835     989     1.06%    1,834,341    1.57%    2,126    2,565    0.99%     4,570,288    1.53%
   2    AK      1,015   1,217     1.30%      255,210    0.22%    2,854    3,385    1.30%       674,995    0.23%
   4    AZ      1,017   1,277     1.36%    2,464,775    2.11%    2,816    3,570    1.37%     6,367,947    2.13%
   5    AR        824     939     1.00%    1,153,428    0.99%    2,185    2,496    0.96%     2,804,822    0.94%
   6    CA      6,600   9,380    10.02%   13,019,578   11.14%   19,673   29,404   11.30%    36,295,313   12.13%
   8    CO      1,574   2,074     2.21%    1,976,010    1.69%    4,224    5,662    2.18%     4,877,134    1.63%
   9    CT      1,623   2,185     2.33%    1,370,773    1.17%    4,393    6,007    2.31%     3,476,086    1.16%
  10    DE      1,116   1,368     1.46%      336,487    0.29%    3,008    3,722    1.43%       862,603    0.29%
  11    DC      1,274   1,711     1.83%      283,771    0.24%    2,620    3,696    1.42%       581,847    0.19%
  12    FL      3,301   3,964     4.23%    7,492,825    6.41%    8,324   10,182    3.91%    18,073,837    6.04%
  13    GA      1,673   1,993     2.13%    3,737,038    3.20%    4,661    5,644    2.17%     9,492,868    3.17%
  15    HA      1,252   1,511     1.61%      431,698    0.37%    3,760    4,537    1.74%     1,267,409    0.42%
  16    ID        815     998     1.07%      565,396    0.48%    2,468    3,121    1.20%     1,501,022    0.50%
  17    IL      2,313   2,959     3.16%    4,908,380    4.20%    6,516    8,570    3.29%    12,688,027    4.24%
  18    IN      1,149   1,299     1.39%    2,490,027    2.13%    3,077    3,545    1.36%     6,262,634    2.09%
  19    IA      1,361   1,612     1.72%    1,238,476    1.06%    3,671    4,527    1.74%     2,969,624    0.99%
  20    KS      1,097   1,300     1.39%    1,110,369    0.95%    2,901    3,441    1.32%     2,721,828    0.91%
  21    KY      1,096   1,267     1.35%    1,715,041    1.47%    2,922    3,366    1.29%     4,206,968    1.41%
  22    LA        680     765     0.82%    1,574,614    1.35%    1,857    2,124    0.82%     4,196,757    1.40%
  23    ME      1,338   1,526     1.63%      546,889    0.47%    3,497    3,941    1.51%     1,312,747    0.44%
  24    MD      1,791   2,271     2.42%    2,090,290    1.79%    4,798    6,204    2.38%     5,565,417    1.86%
  25    MA      1,119   1,365     1.46%    2,462,845    2.11%    3,029    3,855    1.48%     6,339,513    2.12%
  26    MI      1,760   2,033     2.17%    3,967,945    3.39%    4,810    5,529    2.12%     9,926,701    3.32%
  27    MN      1,697   2,126     2.27%    2,096,954    1.79%    4,591    5,816    2.24%     5,190,163    1.74%
  28    MS        748     854     0.91%    1,099,804    0.94%    1,966    2,272    0.87%     2,902,588    0.97%
  29    MO      1,236   1,436     1.53%    2,387,239    2.04%    3,278    3,799    1.46%     5,791,038    1.94%
  30    MT        754     848     0.91%      402,675    0.34%    1,884    2,105    0.81%       939,097    0.31%
  31    NE      1,090   1,337     1.43%      714,414    0.61%    2,924    3,682    1.42%     1,753,474    0.59%
  32    NV      1,206   1,565     1.67%      992,684    0.85%    3,295    4,518    1.74%     2,567,796    0.86%
  33    NH      1,511   1,857     1.98%      513,579    0.44%    4,233    5,181    1.99%     1,314,181    0.44%
  34    NJ      1,614   2,070     2.21%    3,232,691    2.77%    4,527    6,038    2.32%     8,555,736    2.86%
  35    NM        820   1,008     1.08%      778,951    0.67%    2,059    2,511    0.97%     1,946,264    0.65%
  36    NY      3,295   4,133     4.41%    7,484,663    6.40%    8,946   11,387    4.38%    19,062,342    6.37%
  37    NC      1,607   1,903     2.03%    3,716,444    3.18%    4,168    5,054    1.94%     9,183,099    3.07%
  38    ND        913   1,080     1.15%      264,088    0.23%    2,401    2,851    1.10%       614,805    0.21%
  39    OH      2,025   2,375     2.54%    4,523,380    3.87%    5,405    6,408    2.46%    11,300,411    3.78%
  40    OK        952   1,113     1.19%    1,412,585    1.21%    2,560    2,993    1.15%     3,551,011    1.19%
  41    OR      1,026   1,274     1.36%    1,517,902    1.30%    2,718    3,476    1.34%     3,761,928    1.26%
  42    PA      2,272   2,637     2.82%    5,067,441    4.34%    6,086    7,130    2.74%    12,313,327    4.12%
  44    RI      1,185   1,451     1.55%      415,329    0.36%    3,219    4,020    1.55%     1,043,765    0.35%
  45    SC      1,061   1,250     1.33%    1,723,866    1.47%    2,757    3,248    1.25%     4,384,481    1.47%
     46    SD       1,120    1,342     1.43%        330,608     0.28%      2,987      3,630    1.40%            788,285   0.26%
     47    TN       1,031    1,198     1.28%      2,511,303     2.15%      2,720      3,149    1.21%          6,150,384   2.06%
     48    TX       4,126    5,290     5.65%      8,705,751     7.45%     11,798     15,534    5.97%         23,704,369   7.93%
     49    UT         835    1,039     1.11%        902,482     0.77%      2,730      3,512    1.35%          2,657,470   0.89%
     50    VT       1,020    1,214     1.30%        268,767     0.23%      2,565      3,043    1.17%            613,788   0.21%
     51    VA       1,588    1,995     2.13%      2,970,747     2.54%      4,368      5,573    2.14%          7,683,805   2.57%
     53    WA       1,312    1,592     1.70%      2,614,888     2.24%      3,520      4,312    1.66%          6,509,198   2.18%
     54    WV         851      929     0.99%        740,558     0.63%      2,224      2,395    0.92%          1,795,204   0.60%
     55    WI       1,399    1,618     1.73%      2,247,899     1.92%      3,740      4,396    1.69%          5,473,465   1.83%
     56    WY         955    1,120     1.20%        216,917     0.19%      2,545      3,063    1.18%            517,862   0.17%
      Total       75,872    93,657              116,880,816              206,404   260,189               299,105,721

1
    Unadjusted numbers do not include any household replication.
2
    Adjusted numbers include household replicates due to the immigrant imputation and high-income cloning.
                                                     Appendix C
                                   Variables Showing the Greatest Change
                                        In Weighted Sum of Values

                                                                                                  Change from CY 2006 to
                                        CY 2007                                CY2006                    CY 2007
                                                                                                            Pct     Pct
                                                                                                          Change Change
                                                     Wt                                 Wt                 In Wt   In Wt
                         Wt.                       Mean                               Mean         Pct      Num    Mean
                        Sum of     Wt Num           Non-     Wt. Sum     Wt Num        Non-      Change     Non-  of Non-
                        Values     Non-zero         zero     of Values   Non-zero      zero      in Wt.     zero    zero
        Term           (million)    Values         Values    (million)    Values      Values      Sum     Values   Values
LastOwn
WorkersCompIns               85       28,669        2,975         222       21,133      10,498     -62%     36%     -72%
LastBlack
LungMinerBen                 94       12,243        7,651         165       22,878       7,231     -43%    -46%       6%
LastOtherBusiness
SelfEmpIncome            28,368     3,147,980       9,011      46,345     3,531,270     13,124     -39%    -11%     -31%
LastSecondary
SurvivorIncome              487       64,261        7,581         720       70,187      10,252     -32%     -8%     -26%
LastOther
FarmSelfEmployment        6,451     1,360,020       4,743       7,999     1,669,260      4,792     -19%    -19%      -1%
LastOwn
DisabilityIns             2,259      203,093       11,121       2,796      226,526      12,344     -19%    -10%     -10%
LastPrimary
Disability               19,611     1,615,240      12,141      22,469     1,792,970     12,532     -13%    -10%      -3%

LastTotal Disability     19,745     1,615,240      12,224      22,618     1,792,970     12,615     -13%    -10%      -3%
PhoneAvailable               16     9,395,660        1.70          18    10,621,300       1.70     -12%    -12%       0%
LastRental               62,463     9,642,870       6,478      69,693    10,269,800      6,786     -10%     -6%      -5%
RentsOrRoyalties         63,456     9,710,730       6,535      70,680    10,337,200      6,837     -10%     -6%      -4%
Alimony                   5,461       411,218      13,281       4,657       394,625     11,801      17%      4%      13%
MonthlyAlimony            5,461     4,312,750       1,266       4,657     4,137,540      1,126      17%      4%      13%
CHIPCoverage                149    78,651,300        1.90         127    65,042,200       1.95      18%     21%      -3%
Health
CHIPCoverage                149    78,695,600         1.90        127    65,154,100       1.95     18%      21%      -3%
Other
PublicAssistance          2,184      637,353        3,427       1,839      618,860       2,972     19%       3%      15%

Health CoveredOther         0.1      109,974          1.00         0.1      85,451        1.00     29%      29%       0%
LastStateTemp
Disability1                  57       15,626        3,631          16        7,905       2,029    254%      98%      79%
                                                Allocation and Imputation Flags
AllocFlag
ChildRecvSSI               0.01          0.01         1.00        0.04         0.04       1.00     -62%    -62%       0%
AllocFlag
OtherHICoverage            0.80          0.80         1.00        1.24         1.24       1.00     -35%    -35%       0%
AllocFlag
PublicAssistance
Type                       0.08          0.08         1.00        0.11         0.11       1.00     -24%    -24%       0%
MigrationChange
Imputed                    5.06          4.33         1.17        6.19         5.44       1.14     -18%    -20%       3%
    AllocFlag
    ChildRecvSocSec                0.04           0.04        1.00         0.05           0.05        1.00       -15%      -15%   0%
    AllocFlag HasRecvd
    PublicAssist                   5.24           5.24        1.00         6.10           6.10        1.00       -14%      -14%   0%
    AllocFlag
    ChildSupport                   5.90           5.90        1.00         6.83           6.83        1.00       -14%      -14%   0%
    HealthCovers
    OutsideHHImputed              11.48          11.48        1.00        13.01          13.01        1.00       -12%      -12%   0%

    AllocFlag ChildCare            5.58           5.58        1.00         6.26           6.26        1.00       -11%      -11%   0%
    AllocFlag
    FoodStamps                     6.19           6.19        1.00         6.94           6.94        1.00       -11%      -11%   0%
    AllocFlag
    PriorCoverage                 10.56          10.56        1.00        11.80          11.80        1.00       -11%      -11%   0%
    HealthOutsideHH
    Imputed                       36.76          36.76        1.00        40.98          40.98        1.00       -10%      -10%   0%
    Health
    OwnPlanImputed                30.35          30.35        1.00        33.83          33.83        1.00       -10%      -10%   0%
    HealthPrivate
    OutsideHHImputed               3.32           3.32        1.00         3.69           3.69        1.00       -10%      -10%   0%
    AllocFlagGED                  27.23          27.23        1.00        24.28          24.28        1.00        12%       12%   0%
    AllocFlag
    JobReadiness                  27.27          27.27        1.00        24.30          24.30        1.00       12%       12%    0%
    AllocFlag
    CommunityService              27.32          27.32        1.00        24.32          24.32        1.00       12%       12%    0%
    AllocFlag
    JobTraining                   27.33          27.33        1.00        24.32          24.32        1.00       12%       12%    0%
    AllocFlag JobClub             27.36          27.36        1.00        24.34          24.34        1.00       12%       12%    0%
    AllocFlag
    Transportation                17.43          17.43        1.00        15.36          15.36        1.00       13%       13%    0%
    AllocFlagCHIP                  1.79           1.79        1.00         1.57           1.57        1.00       14%       14%    0%

1
    There are so few recipients of LastStateTempDisability that this term is subject to extreme statistical variability.

				
DOCUMENT INFO
Jun Wang Jun Wang Dr
About Some of Those documents come from internet for research purpose,if you have the copyrights of one of them,tell me by mail vixychina@gmail.com.Thank you!