CPS Public Use Microdata

The Census Bureau and BLS publish microdata files from the CPS. The data are available for download and online access, through portals such as MDAT.

Contents

CPS Public Use Microdata

Variable Names

Variable names follow a scheme:

The first letter indicates an analysis level.

Letter	Meaning
P	Individual
H	Household
G	Geographical unit

The second letter indicates a source.

Letter	Meaning
U	Original, unedited
E	Edited
R	Recoded
T	Topcoded (i.e., any value over a threshold is recoded into the threshold)

||W||Weighting (see Weights section)|

\|X	Allocation (see Allocation Flags section)

Missing Values

The following values have indicate a type of missing value:

Value	Meaning
-1	Blank (often out of universe)
-2	Don't know
-3	Refused

Weights

Weights are calculated for and assigned to each record to produce more accurate estimates. There are several weights provided by the Census Bureau, and a use case should be matched to the most appropriate ones.

Weight	Variable Name	Usage
Family weight	`PWFMWGT`	Used for estimates of families
Longitudinal weight	`PWLGWGT`	Used for records that are matched month-to-month
Outgoing rotation weight	`PWORWGT`	Used for estimates making use of only outgoing rotation groups (i.e., months 4 or 8)
Second stage weight	`PWSSWGT`	Used for calibration
Veterans weight	`PWVETWGT`	Used for estimates of veterans and nonveterans
Composited weight	`PWCMPWGT`	Used for estimates of individuals
Household weight	`HWHHWGT`	Used for estimates of households

Weight variables are stored with 4 implied decimal places; they should be divided by 10,000 before using.

For most estimations, the final composited weights are recommended.

When pulling CPS targets for use in weighting, the second stage weights should be used.

Allocation Flags

For some interviews, it is necessary to impute values. These edits are evident in changes from the unedited variable ("U") to the edited variable ("E"). An allocation flag variable ("X") is provided to make the imputation more evident.

Most individual- ("P") and household-level ("H") variables have a corresponding allocation flag. Some recoded ("R") and topcoded ("T") variables do as well.

All allocation variables follow the same scheme:

The first digit indicates the how.

Digit	Meaning
0	No change
1	Changed to some value
2	Changed to an unedited value from a prior interview
3	Changed to an edited value from a prior interview
4	Changed to an allocated value
5	Changed to be blank

The second digit indicates the why, and largely corresponds to the missing values detailed above.

Digit	Meaning
0	Unedited variable was set to a value
1	Unedited variable was blank
2	Unedited variable indicated "don't know"
3	Unedited variable indicated refusal

As an example, if PXSEX=21, then PESEX was blank and imputed from a prior interview.

Edited Universe

Some questions are intended for a subpopulation of respondents. Missingness (i.e., values of -1) is then enforced through allocation recodes. For example, PEEDUCA has an edited universe of PRPERTYP = 2 or 3; it will be set to -1 for all cases where PRPERTYP = 1.

The edited universe for any variable is noted in the data dictionary.

Data

The most commonly used variables, and recommended uses of them, are:

Variable	Usage
`PRPERTYP`	Identifies record as child, adult, or member of US Armed Forces
`PRTAGE`	Age; topcoded at 85
`PESEX`	Sex
`PTDTRACE`	Race
`PEHSPNON`	Hispanic ethnicity
`PEEDUCA`	Educational attainment; edited universe is `PRPERTYP=2 or 3`
`PEAFEVER`	U.S. Armed Forces service history; edited universe is `PRTAGE>=17`
`PEMARITL`	Marital status; edited universe is `PRTAGE>=15`
`PENATVTY`	Foreign-born status
`PEMLR`	Employment status; edited universe is `PRPERTYP=2`
`PRWRKSTAT`	Work status; edited universe is `PEMLR=1 thru 7`
`PEHRUSL1`	Usual hours worked per week at first job;edited universe is `PEMLR=1 or 2`
`PEHRUSL1`	Usual hours worked per week at second job (if applicable)
`PEHRUSLT`	`PEHRUSL1 + PEHRUSL2`
`PEHRWANT`	Seeking full-time work; edited universe is `PEMLR=1 and PEHRUSLT=0 thru 34`
`HEFAMINC`	Household annual income (16 categories); note the high allocation rate
`GESTFIPS`	State
`GTCBSA`	Metropolitan statistical area
`GTCO`	County

These measures are not collected from the entire sample.

Variable	Usage
`PEERNHRO`	Usual hours worked per week if paid an hourly wage
`PTERNHLY`	Hourly wage rate; 2 implied decimal places; topcoded based on the product of this and `PEERNHRO`
`PTERNWA`	`PEERNHRO * PTERNHLY`; 2 implied decimal places
`PRDTOCC1`	Occupation (23 categories) for first job; recode from `PTIO1OCD`
`PRDTOCC2`	Occupation for second job; recode from `PTIO2OCD`
`PRDTIND1`	Industry (52 categories) for first job; recode from `PEIO1ICD`
`PRDTIND2`	Industry for second job; recode from `PEIO2ICD`

These categories are either standard definitions or the prevailing usage of the term:

Category	Meaning
Civilian noninstitutional population	`PRPERTYP=2 and PRTAGE>=16`
Unemployed	`PEMLR=3 or 4`
Employed	`PEMLR=1 or 2`
Labor force	`PEMLR=1 thru 4`
Working part time for economic reasons	`PRWRKSTAT=3 or 6`
White, Non-Hispanic	`PTDTRACE=1 and PEHSPNON=2`
Black, Non-Hispanic	`PTDTRACE=2 and PEHSPNON=2`
AAPI, Non-Hispanic	`PTDTRACE=4 or 5 or 15; and PEHSPNON=2`
Other, Non-Hispanic	`PTDTRACE=3 or 6thru 14 or 16thru 26; and PEHSPNON=2`
Hispanic	`PEHSPNON=1`

UnitedStates/CensusBureau/CurrentPopulationSurvey/PublicUseMicrodata