The basic mechanism of reading in the CPS depends on two included data sets:
-
cps_cols, containing information about column positions in the raw CPS files -
cps_factors, containing information about the factor levels for the raw numeric codes
These contain a set of default data for some commonly-used variables
in the CPS and all of the VRS-specific variables. Across years, there
have been many changes in the questions asked by the CPS, their possible
responses, and where they are located in the fixed-width files that make
up the data. For example, in 1996 the question asking how long a
respondent had lived in their current residence was called
PES6, occupied positions 823-824, and had six valid
responses: less than 1 month, 1-6 months, 7-11 months, 1-2 years, 3-4
years, and 5 years or longer. In 2014, this same question was called
PES8, occupied positions 965-966, and had replaced the
first three categories from 1996 with one response “less than 1 year”.
This illustrates how important it is to provide the correct column
specification and factor levels for any variable you want to read
in.
In an ideal world, we would have provided this information for every variable in the CPS across all years. This quickly becomes time-prohibitive, as there are several hundred unique variables over the 13 years of data. We chose instead to focus on several important demographic variables and all of the voting and registration questions, and have provided full specifications for these most relevant variables.
All that being said, let’s explore how you can add some more
variables to your data. In this example, we’ll read in the 2006-2008 CPS
data with family income as an additional variable. First, we need to
specify which column positions contain the family income variable in
those years - for both years, this is positions 39-40. This is found in
the 2006 and 2008 documentation files, which you can download with
cps_download_docs().
income_cols <- data.frame(
year = c(2006, 2008),
cps_name = "HUFAMINC",
new_name = "FAM_INCOME",
start_pos = 39,
end_pos = 40,
stringsAsFactors = FALSE
)| year | cps_name | new_name | start_pos | end_pos |
|---|---|---|---|---|
| 2006 | HUFAMINC | FAM_INCOME | 39 | 40 |
| 2008 | HUFAMINC | FAM_INCOME | 39 | 40 |
We should then specify which factor levels are needed for those years of data. This is also obtained from the 2006 and 2008 documentation files.
income_factors <- data.frame(
year = c(rep(2006, 16), rep(2008, 16)),
cps_name = "HUFAMINC",
new_name = "FAM_INCOME",
code = c(1:16, 1:16),
value = rep(c("LESS THAN $5,000",
"5,000 TO 7,499",
"7,500 TO 9,999",
"10,000 TO 12,499",
"12,500 TO 14,999",
"15,000 TO 19,999",
"20,000 TO 24,999",
"25,000 TO 29,999",
"30,000 TO 34,999",
"35,000 TO 39,999",
"40,000 TO 49,999",
"50,000 TO 59,999",
"60,000 TO 74,999",
"75,000 TO 99,999",
"100,000 TO 149,999",
"150,000 OR MORE"), 2),
stringsAsFactors = FALSE
)| year | cps_name | new_name | code | value |
|---|---|---|---|---|
| 2006 | HUFAMINC | FAM_INCOME | 1 | LESS THAN $5,000 |
| 2006 | HUFAMINC | FAM_INCOME | 2 | 5,000 TO 7,499 |
| 2006 | HUFAMINC | FAM_INCOME | 3 | 7,500 TO 9,999 |
| 2006 | HUFAMINC | FAM_INCOME | 4 | 10,000 TO 12,499 |
| 2006 | HUFAMINC | FAM_INCOME | 5 | 12,500 TO 14,999 |
| 2006 | HUFAMINC | FAM_INCOME | 6 | 15,000 TO 19,999 |
| 2006 | HUFAMINC | FAM_INCOME | 7 | 20,000 TO 24,999 |
| 2006 | HUFAMINC | FAM_INCOME | 8 | 25,000 TO 29,999 |
| 2006 | HUFAMINC | FAM_INCOME | 9 | 30,000 TO 34,999 |
| 2006 | HUFAMINC | FAM_INCOME | 10 | 35,000 TO 39,999 |
| 2006 | HUFAMINC | FAM_INCOME | 11 | 40,000 TO 49,999 |
| 2006 | HUFAMINC | FAM_INCOME | 12 | 50,000 TO 59,999 |
| 2006 | HUFAMINC | FAM_INCOME | 13 | 60,000 TO 74,999 |
| 2006 | HUFAMINC | FAM_INCOME | 14 | 75,000 TO 99,999 |
| 2006 | HUFAMINC | FAM_INCOME | 15 | 100,000 TO 149,999 |
| 2006 | HUFAMINC | FAM_INCOME | 16 | 150,000 OR MORE |
| 2008 | HUFAMINC | FAM_INCOME | 1 | LESS THAN $5,000 |
| 2008 | HUFAMINC | FAM_INCOME | 2 | 5,000 TO 7,499 |
| 2008 | HUFAMINC | FAM_INCOME | 3 | 7,500 TO 9,999 |
| 2008 | HUFAMINC | FAM_INCOME | 4 | 10,000 TO 12,499 |
| 2008 | HUFAMINC | FAM_INCOME | 5 | 12,500 TO 14,999 |
| 2008 | HUFAMINC | FAM_INCOME | 6 | 15,000 TO 19,999 |
| 2008 | HUFAMINC | FAM_INCOME | 7 | 20,000 TO 24,999 |
| 2008 | HUFAMINC | FAM_INCOME | 8 | 25,000 TO 29,999 |
| 2008 | HUFAMINC | FAM_INCOME | 9 | 30,000 TO 34,999 |
| 2008 | HUFAMINC | FAM_INCOME | 10 | 35,000 TO 39,999 |
| 2008 | HUFAMINC | FAM_INCOME | 11 | 40,000 TO 49,999 |
| 2008 | HUFAMINC | FAM_INCOME | 12 | 50,000 TO 59,999 |
| 2008 | HUFAMINC | FAM_INCOME | 13 | 60,000 TO 74,999 |
| 2008 | HUFAMINC | FAM_INCOME | 14 | 75,000 TO 99,999 |
| 2008 | HUFAMINC | FAM_INCOME | 15 | 100,000 TO 149,999 |
| 2008 | HUFAMINC | FAM_INCOME | 16 | 150,000 OR MORE |
To read income in with our default data, we bind these to the bottom of the included data sets.
Then we can read in the CPS data with our new column specifications and factor it according to the updated factors.
cps_income <- cps_read(years = c(2006, 2008),
dir = here::here("cps_data"),
cols = my_cols) %>%
cps_label(factors = my_factors)
#> Warning in cps_read(years = c(2006, 2008), dir = here::here("cps_data"), : The
#> column names provided by the CPS do not refer to the same question across all
#> years. Be cautious that you are joining columns which correspond across years.
str(cps_income)
#> tibble [304,054 × 18] (S3: tbl_df/tbl/data.frame)
#> $ FILE : Factor w/ 2 levels "cps_nov2006.zip",..: 1 1 1 1 1 1 1 1 1 1 ...
#> $ YEAR : int [1:304054] 2006 2006 2006 2006 2006 2006 2006 2006 2006 2006 ...
#> $ STATE : Factor w/ 51 levels "AL","AK","AZ",..: 1 1 1 1 1 1 1 1 1 1 ...
#> $ AGE : int [1:304054] 35 11 50 78 63 63 37 18 15 8 ...
#> $ SEX : Factor w/ 2 levels "MALE","FEMALE": 2 2 2 1 2 2 2 1 2 1 ...
#> $ EDUCATION : Factor w/ 16 levels "LESS THAN 1ST GRADE",..: 9 NA 10 16 11 4 9 9 4 NA ...
#> $ RACE : Factor w/ 21 levels "WHITE ONLY","BLACK ONLY",..: 1 1 1 1 1 1 2 2 2 2 ...
#> $ HISPANIC : Factor w/ 2 levels "HISPANIC","NON-HIPSANIC": 2 2 2 2 2 2 2 2 2 2 ...
#> $ WEIGHT : num [1:304054] 2411 2666 3254 3647 2754 ...
#> $ VRS_VOTE : Factor w/ 5 levels "YES","NO","DON'T KNOW",..: 1 NA 1 1 1 5 1 2 NA NA ...
#> $ VRS_REG : Factor w/ 5 levels "YES","NO","DON'T KNOW",..: NA NA NA NA NA 5 NA 2 NA NA ...
#> $ VRS_REG_WHYNOT : Factor w/ 12 levels "DID NOT MEET REGISTRATION DEADLINES",..: NA NA NA NA NA NA NA 6 NA NA ...
#> $ VRS_VOTE_WHYNOT : Factor w/ 14 levels "OUT OF TOWN OR AWAY FROM HOME",..: NA NA NA NA NA NA NA NA NA NA ...
#> $ VRS_VOTEMODE_2004toPRESENT: Factor w/ 5 levels "IN PERSON","BY MAIL",..: 1 NA 1 1 1 NA 1 NA NA NA ...
#> $ VRS_VOTEWHEN_2004toPRESENT: Factor w/ 5 levels "ON ELECTION DAY",..: 1 NA 1 1 1 NA 1 NA NA NA ...
#> $ VRS_REG_METHOD : Factor w/ 11 levels "AT A SCHOOL, HOSPITAL, OR ON CAMPUS",..: 5 NA 10 9 9 NA 6 NA NA NA ...
#> $ VRS_RESIDENCE : Factor w/ 9 levels "LESS THAN 1 MONTH",..: 6 NA 4 4 4 9 4 4 NA NA ...
#> $ FAM_INCOME : Factor w/ 16 levels "LESS THAN $5,000",..: 7 7 6 NA NA NA 5 5 5 5 ...One note: the warning from cps_read appears when
join_dfs = TRUE (which is a default). This is intended to
remind the user that variable names change across years, and to urge
caution in only joining the correct columns.
This is an unweighted breakdown of family income responses in 2006 and 2008.
table(cps_income$FAM_INCOME, cps_income$YEAR)| 2006 | 2008 | |
|---|---|---|
| LESS THAN $5,000 | 2800 | 2647 |
| 5,000 TO 7,499 | 2224 | 1963 |
| 7,500 TO 9,999 | 2188 | 2164 |
| 10,000 TO 12,499 | 3447 | 2992 |
| 12,500 TO 14,999 | 3057 | 2815 |
| 15,000 TO 19,999 | 5046 | 4596 |
| 20,000 TO 24,999 | 6352 | 5945 |
| 25,000 TO 29,999 | 6833 | 6192 |
| 30,000 TO 34,999 | 7213 | 6901 |
| 35,000 TO 39,999 | 6662 | 6068 |
| 40,000 TO 49,999 | 10231 | 9951 |
| 50,000 TO 59,999 | 10604 | 9855 |
| 60,000 TO 74,999 | 12607 | 12488 |
| 75,000 TO 99,999 | 14291 | 13838 |
| 100,000 TO 149,999 | 11881 | 12886 |
| 150,000 OR MORE | 8257 | 8936 |